2016Q1 Reports: TACL Journal

From Admin Wiki
Jump to: navigation, search


  • Mark Johnson and Kristina Toutanova have joined Lillian Lee as co-editors-in-chief!
  • Sincere gratitude to our retiring action editor Janyce Wiebe. We are very glad to have the new action editors Phil Blunsom (University of Oxford and Google DeepMind), Chris Quirk (Microsoft Research), Hinrich Schütze (University of Munich), Holger Schwenk (Facebook AI Research), Scott Wen-tau Yih (Microsoft Research), and Luke Zettlemoyer (University of Washington) join, and to have Bo Pang (Google Research) come back as a TACL action editor after a pause.
  • We have expanded the roster of reviewers thanks mainly to Cindy's great efforts.

Policy updates

  • We have a precedent for consequences in case of multiple submission to a conference and TACL, but have not formulated a general policy.
  • Mark Johnson is developing a publication ethics statement. This is an important step in the process of getting the journal indexed by several major bodies.

Biggest pending issues

  • System Hosting.
We are considering move to a different machine hosted somewhere else to improve the reliability of the system.
Professional short-term help by a developer would be very useful if available.

  • Conference interface.
To integrate the handling of TACL papers into the START-based workflow of the conferences, EMNLP 2015 and NAACL 2016 did or will create a special track in START. TACL provides the conference with author contact info, and the authors upload their papers and metadata into the START track. We understand from EMNLP 2015 that this worked well.
Checking for double submissions to TACL and conferences or undeclared re-submission of conference rejected papers: The TACL co-editors and conference PC chairs exchange information on submissions to be able to do this check. We have caught multiple cases of violations of the multiple submission policy.

Selected minor procedure/system updates

  • We have updated the style files for submissions and now require the use of these files.
  • We have updated the reviewer information in the system to improve search for suitable reviewers (thanks to Cindy).

Some statistics

There are currently 22 papers that have not yet received any decision, 7 of which are from the January 2016 round, and thus 15 of which are from the current, Feb 2016 round.

Here is a plot of the history of the number of submissions (meaning "distinct submission IDs", so that a resubmission of a (b) paper isn't included) and the days to first decision for papers that have received a first decision, where the time period is for the lifetime of TACL on the OJS server (we don't have the records for the early days when TACL was hosted at START).

TACL All first decisions.Feb2016.png

Of the 497 papers that received a first decision over this time period, 10% received an (a) = accept with minor revisions encouraged, 22% a (b) = conditional accept, 32% a (c) = reject with encouragement to revise and resubmit, and 35% (d) = reject with 1-year moratorium on resubmission. Of the 52 receiving an (a) on the first go-round, the vast majority were resubmissions of a (c).

For the cases where a reviewer actually submits a review, there is an average of 23 days between the reviewer being notified of a review assignment and submitting a review. However, this statistic doesn't factor cases where a reviewer never completed a review or had to be replaced.

Misc. notes:

  • Submissions for which some technical problem required resubmission are not double-counted. There have been some 40 papers over the past year where the submission violated formatting rules (often A4 was mistakenly submitted instead of letter). This takes up more manual processing and emailing time than seems right,.
  • Days to first decision is approximated in several ways, such as that the date of submission is rounded to the nearest submission deadline.

Number of papers published to date: 126

Number published in 2015: 43.

Roughly 25-30 currently eligible for presentation at this year's conferences, so far.