Breaking News: Changing Attitudes and Practices

Abstract

Standard practice in our field has been to announce research results at our annual conference or one of its affiliated meetings such as EMNLP or at the biennial COLING conference. This year, by its submission deadline of 23 January 2007, the ACL Program Committee had received 588 main conference submissions, plus another 52 submissions to the Student Research Workshop, for a total of 640 papers. Similarly, by its subsequent deadline of 26 March 2007, the EMNLP Program Committee had received 398 submissions (excluding ones that were withdrawn or rejected without review). It was estimated that about a third of these (say 130) were the same or minor variations of papers submitted to ACL conference.1 With over 900 separate submissions, one might wonder if all breakthroughs in our field are really made in late fall or winter, just in time for these deadlines. If they’re not, why is it that these deadlines seem to define when new results are announced? Is there no credit to be gained from really being the first to publish some new method or theory or some clever take on an old one? Or are there no places to publish that will guarantee catching the field’s immediate attention (our equivalent of Science, Nature, or YouTube)? In short, why the veritable flood of words crashing up against conference deadlines and the veritable trickle reaching the editorial offices of the significant (and still growing) number of CL/NLP-related journals. A choice is clearly being made by researchers in the field, but is it one that should be encouraged? Could change bring about some better situation? Although our journals and conferences are well-respected (and the latter are also great fun and a major contributor to our sense of community), frustration with both has been heating up over the last year or so, and clear calls for change are in the air. The following is a summary of what I myself believe or have heard others claim to believe, along with some suggestions for possible solutions. I am indebted to discussions with Aravind Joshi, Mark Steedman, Lauri Karttunen, Julia Hockenmeier, Annie Zaenen (co-Editor-in-Chief of Linguistic Issues in Language Technology), John Tait (Executive Editor of Journal of Natural Language Engineering), Kam-Fai Wong and Jun’ichi Tsujii (co-Editors-in-Chief of ACM Transactions on Asian Language Information Processing), Shalom Lappin (co-Editor-in-Chief of Research on Language and Computation), and Robert Dale (Editor-in-Chief of this journal, Computational Linguistics), as well as the many comments I have read at the Natural Language Processing blog