Is it acceptable to use non-English references in a computer science paper?

Question

I am currently reviewing an interdisciplinary computer science paper which is heavily based on previous works in logistics and process engineering that have been unfortunately published only in German (18 out of total 30 references). The overall idea is sound, but as the logistics structures are in some sense a basis of the paper, I have a strong inclination to reject the manuscript as the research cannot be followed and reconstructed by a researcher that does not understand German.

On top of that, virtually none of them are available online, or otherwise easily accessible.

I was going to hit the "Reject" button just a few moments ago, but I still have my doubts: I know that in humanities and also in mathematics people often cite publications written in other languages, but I have not yet seen something like that in computer science. Also, if it were one or two references, but 60%? The journal reviewing policy does not help me in this case.

I was thinking about proposing the following workarounds:

• Extending the paper - this is probably not an option as the journal imposes quite a strong page limit, or
• A supplementary report - the vital parts of the non-English text could be made available as an Technical Report or an on-line publication (I was thinking about arXiv) and this can be referenced in the manuscript.

Is it fair to reject it? Has anyone experienced something similar? What was your decision?

Answer 1

Having many non-English references is absolutely no valid reason to reject a computer science paper.

If a reader wants to follow up on the references, they can learn some German, find a translator, etc. But this is the responsibility of the reader, not the author.

Answer 2

I tend to agree with @Arno's answer that citing non-English sources is fine. But that is actually beside the point, since as far as OP's situation is concerned, I think he has much less of a dilemma than he thinks.

The reason is that whether to publish a paper that references many German-language sources should be an editorial decision. As a reviewer, OP should check that the paper is correct (including looking up the references to the extent that's necessary, which OP can do since he speaks German), and that it is novel and lives up to the standards of the journal. If those conditions hold, he should recommend acceptance. He can also point out in his report the potential issue with non-German readers not being able to fully understand the theoretical background the paper makes use of. It would then be up to the editor to decide how to handle the situation, depending on their philosophy and the policies of the journal (which may vary, for example a journal published in Germany might have a different view of such issues than one published in the United States).

To summarize, the question of whether papers like this should be published is a very interesting one, but from a practical point of view, I don't think OP really needs to concern himself with it.

Answer 3

Your only responsibility as a reviewer is to check that the article is scientific, relevant and original enough. When references are valid, the language should not matter, especially if - as you say - you can check them because you personally do know German.

In other words: I too am shocked that this is even an issue.

Answer 4

See it the other way round. Many German papers have English references, although not all readers will be fluent in English. If you cite from a foreign book, eg. chinese it is helpful for the reader to give a (rough) translation.

If you reference to a paper from an 'exotic' language to translate the tile and give a very short synopsis of the content. Because it is helpful to know whether the reference is a paper on a specific problem or giving an overview over another topic.

If you are restricted in length in a journal, create a small website, were you provide the German original sources and make it readable to the english reader, by letting out the German citations.

Is the paper still valid with many foreign references? As a informed reader I would ask whether the reference given are available in the language of the paper. E.g. referencing to a standard algorithm in a german book, when the same algorithm is described in many english books. If this is the fact I would be more alert concerning the validity of the paper.

However brilliant papers have been written in China, behind the iron curtain and so on, without any reference to english sources.

I'd be happy to read a such brilliant work

Answer 5

From a scientific/scholarly point of view, there is absolutely nothing wrong with non-English references which date from a time when English wasn't quite as dominant yet, or which come from a field where it still isn't.

More recent non-English references can be a red flag but they are still no reason to reject a paper. They are a red flag because it's just not normal any more, and so people who publish in a language other than English anyway are likely to be very inexperienced, very old, very eccentric, or trying to hide plagiarism from English sources (possibly self-plagiarism). All of these factors are reasons to have a close look, which requires being able to read the references. To some extent these factors apply even when the references are from a technical field that hasn't standardised on English yet, as this makes it likely that the field itself has low scientific standards.

In any case, however, if for some reason you cannot really review a paper, then you should refuse doing so. Your inability to read some key references would certainly be one reason to do so. Or you could delegate the review (or just checking the references) to someone who can read the key references that you can't.

Although this was historically not done, I guess nowadays one should be very careful when citing non-English sources. Ideally, the author should be very explicit about the major claims taken from such a source. This way most readers can simply take it on faith that the source actually supports the claim, and those readers who can read the source (hopefully including a reviewer!) will know if it doesn't, without having to fully understand how it is used.

Answer 6

I'm not sure how computer science publications work, but having used quite a few non-English papers myself I have the following suggestions.

Almost every paper adds a small increment to the specific subject. Usually, authors can easily find one or several English language papers connected to the non-English paper and cite those instead to support their arguments.

In case the authors cite data from a non-English article, it quite acceptable. In this case they must reference strictly numbers that can be looked up in the article by a reader who doesn't understand the language. If there is a need to cite methods or interpretation of the data, then they should be translated and briefly summarized in English.

In any case, the burden of translating and verifying foreign texts should not fall on the reader of a journal.

Answer 7

I like your idea to propose the mandatory provision of a supplementary report.

At the cost of some extra effort for the authors, this would be a good compromise between the interests of the authors (getting a fair evaluation of their actual work) and those of the readers (being able to reconstruct and extend the research).

An included benefit for the authors would be the possibility to disseminate the underlying ideas to a wider audience.

Answer 8

Rejecting a paper because you do not understand the language doesn't sound like a good excuse and people will get frustrated.

On the other hand, since it is not some standard and well known/used publication that has been scrutinized and/or translated, and it cannot even be found online it is up to the authors to make every possible effort and convince the readership that their techniques and relevant results are sound.

I suggest to give them the chance to enhance and update their manuscript with all the necessary details in order to judge the soundness of the paper. Failing to do so would (and should and a fair warning) lead to a regretful rejection.

Answer 9

I would not go for a reject for this reason.

As a reviewer, wishing that authors provide more background from foreign language papers is acceptable. I would do so in this case.

Like, in normal case it is straightforward to assume that the reader can inform herself on previous techniques from earlier papers and you present the state of the art more to highlight what you have done differently.

Here it is not like this. So, the authors might be required to present a better/broader overview of the related work and state of the art. As a reviewer I would basically ask them to retell the essence of the foreign-language papers, so their paper is understandable and makes sense even if the reader cannot read the foreign language references.

As a side note: where it mattered, translations were published. You paper authors might not be aware of them, but they might be out there. I know that some Russian math journals were routinely translated in English during the Cold war. It's another issue how to cite such works in an English paper...

Answer 10

I would recommend rejecting any paper that is incomprehensible to somebody in the field without following its references.

The references not being in English isn't the problem. Leaning on the references too much is.

Answer 11

What is the purpose of references? Certainly allowing the average reader to follow the research is a part of it, but the more important part is giving appropriate credit to where the authors got their ideas, which may well include papers in the authors’ native language or which happened to available to them but not the wider internet.

In itself, definitely not a reason to reject.

That said, the paper should stand reasonably well on its own - the text should include sufficient summary that only those particularly interested in the details need look for translated versions of the citations. Also, a qualified reviewer should be able to confirm that the cited paper says what the authors claim it does.

Answer 12

There is no simple answer to this question: either you are able to actually judge on the quality of the cited material, or you are not in a position to be the reviewer for the particular paper.

Citing inaccessible material is simply slightly negatively affecting your evaluation, that's it. You should definitely reject if the author cites a simple corollary of a theorem to be found in a privately circulated memoir of the Slovenian Philological Society, 1883. This citation cannot be serious in a CS paper unless the paper is doing research on citations or invalid proofs. On the contrary, citing Gauss, Leibniz, Euler, Grothendieck, Kantorovich, Fichtenholz, or Kolmogorov is actually not a bad sign by itself. In general, you never know before you look.

In the particular case, since you happen to be fluent in German, there is less excuse for not being able to produce a good review.