Professor of Computational Linguistics
Joakim Nivre is Professor of Computational Linguistics at Uppsala University. He holds a Ph.D. in Linguistics from the University of Gothenburg and a Ph.D. in Computer Science from Växjö University. His research focuses on data-driven approaches to morphosyntactic analysis and linguistic resources needed for these tasks. He co-authored the standard textbook on dependency parsing and led the development of MaltParser. He served as secretary of EACL (2009-2012), secretary/president of SIGNLL (2007-2011), editorial board member for CL (2010-2012), action editor for TACL (2012-), and in major organizing roles for over 20 international conferences.
Computational linguistics is a thriving scientific field and ACL is flourishing with it. As executive committee member, my main ambition would be to sustain the positive development resulting from wise strategic decisions in the past, and I believe my broad experience from different ACL activities puts me in a position to do so. However, we must constantly face new challenges, and I think there are two areas in particular that demand our attention.
Publishing and reviewing: By tradition, most of our work is published in conference proceedings rather than journals, which can be a disadvantage when we compete with researchers from other fields for grants or promotion. In addition, the traditional conference reviewing model is showing signs of strain, with growing reviewer loads and declining review quality as a consequence, which also points to the need for more journal publications. The creation of TACL has been a successful first step in this direction, but how do we scale it up? Can we maintain the quality when volume grows? Do we need more journals for conference-length articles? Do we need a different reviewing model for our conferences? I don't know the answers, but I am convinced that we need to take these questions seriously.
Scientific methodology: As the amount of empirical research in computational linguistics has grown, more emphasis has been put on scientific principles such as reproducibility. This has led to a number of good initiatives, in particular to encourage researchers to release code and data, work that needs to be continued. However, the question of scientific methodology goes beyond the sharing of code and data. Researchers from neighboring fields, when reading our papers, sometimes point out that it is hard to see what the research questions are, let alone what specific hypotheses are being tested in relation to these questions. I believe an important part of becoming a mature science is to develop a greater awareness of scientific methodology and better articulate the ways in which we contribute to the advancement of human knowledge. Again, I don't know exactly how, but I am willing to listen and work with you all on this.