You are here: Computational Linguistics > What the Editor Does


What This Page is About

These notes were written by the then-editor in April 2012, when nominations were being sought for a new editor. They are intended to provide some background for those interested in taking over the role; but they may be of more general interest. They reflect one editor's practice, based on ten years' experience in the role; I have no idea how similar or different this practice is to that of other editors, and it's surely the case that there are better ways of doing some of the things below. But this is what worked for me.

The topics covered here are primarily a catalog of the day-to-day responsibilities of the Editor. The Editor also has somewhat self-chosen higher-level responsibilities in regard to planning and managing the journal's future development, but these are harder to characterize. Some of these surface parenthetically below.

Robert Dale, 27th April 2012

Handling Submissions

  1. Submissions are received online by our instantiation of OJS (the Online Journal System), and the bulk of communications and processing in regard to a submission are handled through the site. Once a submission is received, it is checked for completeness by CL's Editorial Assistant. At this point, if she suspects that the submitted article may be inappropriate for CL—Suzy, our Editorial Assistant, has a keen eye for spotting submissions of this kind—she alerts me to this fact, and I take a look at the article to form a view. This usually doesn't take very long—sometimes it's immediately obvious that the article is out of scope—but sometimes a closer look is required. I have rarely taken more than 30 minutes to make such a decision, and it usually takes considerably less. Common reasons for considering an article to be inappropriate for CL at this point in the process (what some journals refer to as a 'desk rejection') are its subject matter and its length, the latter often being a good diagnostic for the depth of substance of the work. Closer inspection of articles that seem within-scope for the journal often reveals they are thin on references to relevant related work, suggesting an unfamiliarity with the field, and this is often borne out by introductory material that reads more like an undergraduate essay than a competent scholarly contribution. The number of inappropriate submissions we receive has increased significantly in the last year or two; they often look like conference papers or articles formatted either generically or for some other journal, which may suggest that 'submission spam' is on the rise. If I deem the article inappropriate for CL, a polite rejection mail is sent via OJS, with a short elaboration of the reasons for rejection, and a suggestion for alternative venues if that seems appropriate.

  2. The above is not the norm, since most submissions pass Suzy's first appropriateness test. Typically the first time I hear of a submission's existence is when I am cc'ed on the email Suzy sends to the editorial board seeking interested reviewers (this email contains the authors, title and abstract, but not the full paper). I don't do anything at this point other than file the mail in a new folder I create for that article (I'm a folder person rather than a tag person when it comes to email).

  3. Shortly thereafter, Suzy sends me an email that collates the essence of the responses to the call for reviewers. This is conveniently structured into three parts, all constructed from material we explicitly request in the mail sent to the board members: brief comments on the appropriateness of the article for CL (based on the content of the title and abstract), an indication of which board members would like to review it, and a list of potential reviewers who are not on the editorial board. We rely on external reviewers a lot, both to lighten the load placed on editorial board members, and to access specialist expertise that might not be present on the editorial board.

  4. I try to process these emails very promptly, to reduce the time lags that we have some control over. I send back to Suzy an edited copy of the email with a rank ordering over the candidate list; we aim for three or four reviewers per paper, but I rank order more than this (sometimes as many as 10) so Suzy can quickly respond to refusals. I generally try to ensure that we use at least one external reviewer for each article; this brings fresh blood into the system and also lets me 'try out' potential new editorial board members. At the same time I try to ensure we have at least one editorial board member involved, to provide some kind of calibration across articles. I avoid including in the ranking anyone who I consider to be an unreliable or unhelpful reviewer.

  5. The next I hear of an article is when all the reviews are in. OJS automatically highlights articles for which this is the case, but Suzy sends me an email to alert me, along with information indicating the submission date and any previous decision dates in the case of revised submissions (I use that information when I send acceptance letters; see below).

  6. I look at the reviews in OJS. Over the years we have revised our review form to a format that works well for me, but is not universally loved by reviewers. It imposes a fair amount of structure on the review, which makes it easier for me to compare the reviews and extract essential information. I like to think it also imposes some discipline on the reviewers.

  7. The easiest case to deal with is when all the reviewers recommend rejection. The structure of the review form makes it fairly straightforward to determine what the major failings of the paper are. If all the reviewers appear to have compatible views here, and a reading of the reviews doesn't raise any red flags, I will immediately send a rejection to the author (via OJS), incorporating all the reviews. Sometimes I will add some summary comments if I think these will be helpful in understanding the reviews, but the reviews usually speak for themselves. I then send a message (again via OJS) to the reviewers indicating that I have made the reject decision, so they know what is happening; they also get a copy of all the reviews so they know what the other reviewers thought.

  8. In some cases of unanimous rejection by the reviewers, I get a sense of disagreement between the reviewers as to what they think the fundamental problems are. In such cases I will first initiate discussion with the reviewers (again via OJS) before making a final decision; often this is no more than a message that says 'I'm sending a rejection in 72 hours; please read the attached reviews and alert me to anything that someone else says that you disagree with.' The degree of circumspection I express in this message varies depending on the comfort I feel with the views expressed in the reviews. I handle these kinds of cases in around 15 minutes, although more worrisome cases can take longer; yet again, the structured review form helps immensely.

  9. The most common outcome from a review round is a mixture of 'revise and resubmit', 'basically acceptable', and 'reject' recommendations. Sometimes (but rarely) all three appear. The amount of disagreement determines the amount of work required, and again the structured form helps a lot. My aim is to achieve some kind of consensus view via discussion amongst the reviewers, particularly if the outcome is likely to be a revise and resubmit decision. The entire reviews are shared amongst all the reviewers (including a section which is labelled 'for the editor and other reviewers only'). I think it's very important to have clear requirements for the author, so I try to get agreement (at a high level) as to the most important things the author needs to address in order to make the paper publishable. I read the reviews, try to determine what the most important points are, then initiate a discussion with the reviewers to establish agreement; by doing this I aim to avoid situations where one reviewer subsequently raises new show-stopping objections when the revised version of the paper comes in. The points agreed on will generally be incorporated into my decision letter as a set of bullet points that precede the full reviews. The initiation of the initial discussion here, since it requires reading the reviews and attempting to synthesize their key points (once more facilitated by the structured review form), typically takes 15–30 minutes, and only occasionally longer. Of course, the subsequent discussion with the reviewers can take a variable number of iterations, and a varying amount of time. In the early days, I tried to read those papers where there appeared to be significant disagreement amongst the reviewers, but I think this is simply not feasible if you want to have a life; and besides, I found that in very many instances I was not qualified to offer a view on the subtler points being argued in any case.

  10. A common point of disagreement amongst the reviewers is just how much is required to make something publishable. Not surprisingly, different reviewers have different standards, and so I will often inject into the discussion a view that attempts to calibrate based on my experience of our publication history. I cannot claim to achieve consistency here; it's very easy to be swayed by local data (what the reviews in front of me say) in opposition to global data (a dim recollection of what we did last time we had a paper with something like this weakness).

  11. Once consensus (or something like it; of course, consensus is not always possible) has been reached, I send the author a decision letter indicating the outcome. In the case of revise and resubmit decisions, this includes a request for a detailed covering letter indicating how they have responded to each of the reviewers' comments; this covering letter is provided to the reviewers along with the revised submission when it arrives.

  12. In my decision letter, I also indicate a target date for the revised version of the paper, based on the extent of the changes required (typical values here are 1, 2, 3 or 6 months); the authors are invited to negotiate a different target date with Suzy if they have difficulties with the one proposed. We initiated this process because we realised that a large amount of the publication delay journal authors so often complain about arises from them taking so long to revise their papers.

  13. The typical sequence of decisions for papers we receive is first a 'revise and resubmit' decision, and then a 'basically acceptable' decision. This latter is not the highest possible accolade: that's 'acceptable as is'. The 'basically acceptable' decision means that there are still some minor things to be fixed. Depending on the circumstances, this may mean that the paper will go back to the reviewers yet again, or that the authors will be trusted to make the requested changes, with the editor simply checking the covering letter to ensure that all the points raised appear to have been addressed.

  14. Once the paper has reached the point of acceptance, I send the authors an appropriate email including the relevant submission and decision dates for incorporation into their final version (we print this data on the first page of the article). This is the last I hear of a paper until the proofs arrive from MIT Press; more on that below.

  15. Of course there are always some deviations from the norm described above. One or two rounds of revision are the most common; sometimes there are more than this, and in a few uncomfortable cases we have called in an additional reviewer after a number of rounds of revision. Only very occasionally do we get unanimous acceptance of a paper based on its first submission.

Other Responsibilities

The activities around submissions, as outlined above, are probably the most obvious responsibility of the editor, but there are some other things that the editor does:

  1. Choosing Content for Issues: When we were a print journal, our limit of 164 pages per issue meant that articles had to selected from the queue for publication in the next issue; the Editorial Assistant would compile a list with page counts and the Editor would make a selection. Now that we are electronic only, we generally publish papers on an as-ready basis. However, there are still per-page budgetary consequences that arise from the Press's production process, so when we have a glut of papers waiting, we put the brake on a bit by holding some papers back for the subsequent issue.

  2. Handling Proofs: The Editorial Assistant manages the collation of material for submission to MIT Press in line with a schedule provided by the Press, so that happens in the background as far as the Editor is concerned. The Editor next gets involved when the proofs for an issue (we are still issue-based) come back from the Press; these are emailed to the contact authors and cc'ed to me, and the authors are requested to send any changes they want to make to me. In earlier times, authors sent their corrections directly to the Press, but I often proofread the articles too (so this is usually the first time I actually read papers that have been submitted), and I often have a longer list of corrections than the author does; so I act as a collection point for changes. I suspect many editors would consider this degree of involvement over the top.

  3. Selecting the Editorial Board: We appoint 6–8 new editorial board members each year, with a term of three years. I approach the selection task by asking the editorial board for suggestions, resulting in a list that I filter and order based on: the research areas of those proposed; gender and geographical balance; and my own experience of the nominees' reviewing ability and reliability. Once I have a near final list I solicit confidential commentary on that list: existing board members often have private knowledge that is important for decision-making here. Of course one has to be conscious of the scope for personal disagreements interfering here, but to my knowledge this is extremely rare. Once the new members have been selected and invited, and they have all accepted (sometimes we get turned down, so I always have a couple of reserve candidates), I send out a collection of emails announcing the fact to the appropriate constituencies in the appropriate order (the existing board, the new members, the ACL Exec, and the ACL membership). Finally, of course, the outgoing members are thanked for their service.

  4. Reporting to the ACL Executive: The CL Editor is an ex-officio member of the ACL Executive, and provides reports to the two meetings of the Executive held each year (typically a teleconference in January or February each year, and a face to face meeting of the Executive at the ACL conference each year). These reports consist of statistical data compiled by the Editorial Assistant and textual commentary on important aspects of the journal's management and development; previous reports are available from the ACL website.

  5. Hosting Meetings of the Editorial Board: By tradition, the Editor hosts a meeting of the Editorial Board at the annual ACL conference, provided of course that the Editor is in attendance. These are usually one to one-and-a-half hours in length, and are typically scheduled at lunchtime or breakfast; the ACL kindly pays for catering for this event. I have tended to use these meetings to encourage open discussion of more strategic issues regarding the journal's future, but there are no particular rules or requirements here; really it's an opportunity to thank the board members for their work, and to encourage a sense of community.

  6. Handling Special Issue and Survey Article Proposals: We have particular procedures for these, which aim to reduce needless effort on the part of the proposers; see elsewhere on this website for the processes used.

  7. Hiring Editorial Assistants: Having a competent Editorial Assistant working on the journal is essential to things working smoothly, and you really could not do it without one. The ACL pays for one day per week of assistance; so far, this has been sufficient to carry out the tasks required, and in fact since we moved to OJS (our previous submissions management database was a spreadsheet), it has freed up some time that has allowed the present Editorial Assistant to carry out other tasks such as updating and expanding our website, and revising our LaTeX macros. I have been fortunate in the extreme to have been able to exploit a series of able PhD students who have taken this on as a part-time job.

  8. Liaising with the ACL Business Manager: Annually, the Editor estimates the costs for editorial assistance and any other journal-related expenditure, agrees the budget with the ACL Treasurer, and organizes administration of the funds with the Business Manager. Currently the Editorial Assistant bills the ACL periodically, so the Editor approves these invoices.

  9. Liaising with MIT Press: There are a variety of interactions with the Press throughout the year. Many of these concern relatively routine matters such as website updates, scheduling and budget information, impact factor data, and so on, but there are also more strategic issues that arise from time to time. For example, at the time of writing there are ongoing discussions between the Editor and the Press in regard to (a) finding LaTeX-aware copyeditors, as a way of reducing production costs, and (b) moving to full XML publishing.

All the above costs the present Editor an average of around half-a-day in a typical week (the CL Editor is not a paid position, in case that's not obvious). This can blow out somewhat if there are proofs with lots of problems: this is increasingly an issue as we receive more material from authors who are not native speakers of English, since the level of copy-editing we pay for doesn't stretch to fixing significant problems in the English. It helps if you are happy spending weekend mornings in your local cafe adding comments and corrections to proofs.

Anything Above Not Clear?

Make it better for the next user: let us know by sending us an email.

Please send comments or queries about this web site to
Last Modified: 14th June 2013