I am a student of linguistics who works on Marathi. A lot of literature on this topic is available in Marathi language which uses the Devanagari script, but in the academic literature, I have seen many researchers transliterating the original title, author name in the Latin script (with English conventions) e.g.,

Damle, Μ. K. 1970 Shastriya Marathi VyaakraN [A scientific grammar of Marathi]. K. S. Arjunwadkar. Pune, India: Deshmukh & Company.


दामले, मो. के 1970 शास्त्रीय मराठी व्याकरण [A scientific grammar of Marathi]. कृ. श्री. अर्जुनवाडकर. Pune, India: Deshmukh & Company.

I personally don't like this practice. Here's why.

  1. It's not about the pronunciation at all. If it really was, people would have got uncomfortable with Jean, a French name too as it should have been John as per the English conventions. I always used to mispronounce it as jeen from jeans without the s ([ʤiːn] in IPA). This clearly means that people are just not happy to see a script other than Latin in academic writing which I feel is very unjust in a multicultural and multilingual world.
  2. If I am citing a work in a language other than English anyways the reader will have to read it in that language. If they don't know the language, they will probably need a translator for accessing the work. So anyways just transliterating the title is not going to help them in any ways.
  3. If a reader knows that particular language, then it is even annoying for them because there is no "universal" transliteration scheme available. Different researchers follow different conventions without even giving the rules for parsing them. It obstructs the reading of the person who is actually able to read it without any aid and comprehend the language in which the source is written on the cost of the so called "comfort/convenience" of a person who is not able to (or probably going to) read the actual source.

Can we just start getting comfortable with scripts other than Latin in the bibliography?

  • 72
    I am confident that you would very easily be able to answer your own question if, instead of focusing on the case of a script that you can read and on your frustration with having to accommodate readers who cannot read it, you focused instead on a script that you cannot read, such as (I am assuming) Chinese or Japanese characters. I do not for a single second believe that you do not see what the benefit is of referring to, say, Genji Monogatari as opposed to 源氏物語, or to the book Sanaaq by Mitiarjuk Nappaaluk, as opposed to the book ᓴᓈᖅ by ᒥᑎᐊᕐᔪᒃ ᓇᑉᐹᓗᒃ. Apr 23 at 16:17
  • 2
    Can you explain "It's not about the pronunciation?" This seems like an argument in search of a critique Apr 24 at 19:33
  • 6
    I'm in a very different field, so take what I say with a pinch of salt, but when I check a reference, pretty much the first thing I do is read the title to get a sense of what the paper is about and whether it's relevant to me. Only then do I decide whether to look the paper up. If I can't read the title (or if, as in some physics journals, it's not included in the reference) then I'm deprived of the ability to do that, and have to do a lot of extra steps just to figure out if it's a relevant paper or not. Hence I would advise including translations as a courtesy to the reader.
    – N. Virgo
    Apr 25 at 1:56
  • 1
    @AdamPřenosil when the subject of the paper is a language written in that script (as is the case here) it becomes somewhat more reasonable to expect readers of the paper to be familiar with it
    – Tristan
    Apr 25 at 9:24
  • 6
    @N.Virgo but the question is about transliteration, not translation. The example provided by OP still provides a translation even in the non-transliteration version.
    – Muzer
    Apr 26 at 11:56

10 Answers 10


The first consideration is always going to be the requirements of the journal you publish in. But given you're asking this question, I will assume the journal's style guide is silent on that point.

Reading the other answers, it occured to me that even when the reference is in English, it's often true that neither the author names, not the title is very helpful in finding the reference. If I'm looking for a paper called "Search for the Decay K+→π+νν¯¯¯" by "J.Smith et.al.", I'd really prefer to have an arxiv ID or DOI. There are too many "J.Smith" authors, and the title contains lots of characters that will be rendered variably by different software. Even pasting it in here is enough to mess it up (I know I could write it in LaTeX, but I'm making a point about variable rendering).

I think the provided 3 conditions are satisfied, the reference is fine without transliteration.

  1. The original wasn't transliterated. If the original paper/book (or it's author list) was transliterated, then you will make it harder even for a reader fluent in both languages if you refuse to follow suite. So in that case, the same transliteration as the original would be required. This would be unchanged even if different references used different transliteration schemes.
  2. The inline reference contains only characters native/comprehensible in the the language of the paper. For example in physics an inline reference is a number in square brackets, the third reference would be [3]. This matters becuase if the reader struggles to parse the inline reference, they will struggle to find the full reference in the reference section. This also allows the reader to make a verbal statement about the reference when discussing the paper without needing both languages. People often read papers in groups, so that's important.
  3. Your reference section must contain a DOI, URL or other widely recognised unique ID for every reference. This is by far the easiest way to find a referenced publication even if you can read the whole long reference. Author name and year is tedious, most authors publish many things in one year. Sometimes a title and author name is ok, but in your case the title likely contains characters that were written with LaTeX macros (or just are not ascii), and so are ommited, or rendered variably in many databases. This is a problem with maths in paper titles too. So even without another language involved, a DOI or url is normally the fastest way to get the right reference. These things are language agnostic, so you sidestep the question of transliteration.

If someone retrieves a reference and is really confused as to its relevance, then they can (with a little difficulty) make a visual comparison of the title you gave and the reference title. Sometimes we mess up when copying a URL, so that check can be needed. But it's unusual, so I wouldn't see that as too onerous.

Edit; @DikranMarsupial raises the point in the comments that an authors name after transliteration is more memorable to those who only know Latin scripts. As such, transliteration may increase the reputation of the author, and in turn improve understanding of the academic strength of their culture. The drawback being that some people don't want their name transliterated. At least for living authors, we should probably follow their example (as in point 1.). For dead authors, it's difficult to guess what they would opt for if they knew the culture of today, and I'm not sure there is a right answer to this dilemma.


You’re right, it’s not about pronunciation at all. It’s about the reader being able to actually do things with the citation, such as remember or copy the name so that they can look it up later. I can’t type or write using devanagari (so my only option to copy your citation would be to copy and paste it), I can’t remember things written using devanagari (so even if I could type devanagari characters, I wouldn’t be able to copy down the citation later). In other words, even though I would need to find a translation of the paper to read it, I would have to take extra steps just to copy down the reference in many cases, which makes it even less likely I’ll actually benefit from it. Adding to this though, recognition is important. I can remember the transliterated name of the author well enough to be able to easily recognize other works in the citations by the same author, but the devanagari would require me double checking a couple of times to confirm that they are the same name.

Put differently, consider this from the perspective of scripts you cannot read. For example, who is Ива́н Петро́вич Па́влов? Or Ἀριστοτέλης? Or محمد بن موسی خوارزمی? Or 孫子? Of those, I had to copy all of them but Ἀριστοτέλης (and that one I can type only coincidentally, I don’t actually speak or really read Greek, either ancient or modern). All are major names in their respective fields of study, but I’d be willing to bet you have to look up three out of the four just to see who they actually are. In contrast, listing the names as Ivan Petrovich Pavlov, Aristotle, Muhammad ibn Musa al-Khwarizmi, and Sun Tzu makes them readily recognizable (well, three out of four, not all of them are exactly household names in all parts of the world.).

  • 2
    But by using Aristotle, you are already anglicizing the name, not just transcribing it. The Cyrillic for Pavlov looks a bit less complicated without those unnecessary diactitics - those are only used for learners in dictionaries and textbooks (even though Google does also show hits for "Iván Petróvich Pavlov". Apr 24 at 10:57
  • 7
    @VladimirFГероямслава If we want to be picky about transliteration, then strictly speaking Sun Tzu is wrong also (correct Pinyin would be Sūnzǐ). In both cases, I went with the well known names among native English speakers with the intent of ensuring recognition. Also, point taken about Pavlov, I did not know the diacritics are optional (I’ve never been as well versed in usage of Cyrillic in any languages, and in this case just copied the Russian form of the name directly from Wikipedia) Apr 24 at 12:05
  • 3
    A nitpick (and again shows how culture can differ), al-Khawarizmi is more-or-less household name in where I live. Pavlov, not so much. Apr 24 at 16:48
  • 2
    @AustinHemmelgarn Adding another nitpick, Sun Tzu is a legitimate transliteration, it’s just the older, less common Wade-Giles system.
    – H Huang
    Apr 26 at 3:05

Unless there is a compelling reason to the contrary, academic work is written based on the language of the journal and its readers. So for an academic journal that is written in English, we would not usually presume that the reader has knowledge of another language. In certain countries where the general population is bilingual there may be journals that accomodate two languages, allowing some wiggle-room. Similarly in cases where the subject of the journal is linguistics, or some similar field, you might find that the average reader is multilingual and in some cases it may be reasonable to assume knowledge of a secondary language. While this is possible in some contexts, it is relatively unusual, particularly in English-language journals.

Now, suppose we are dealing with an English-language journal where it is not reasonable to expect the reader to speak any secondary language. In this common case, while an English translation/ transliteration (hereafter just "translation" for brevity) is often imperfect, it will generally impart more information than the untranslated language to a reader who knows only English. For a reader who only knows English, the character string शास्त्रीय मराठी व्याकरण is essentially just gobbledegook, whereas its English translation (however imperfect) at least gives some understanding of the words in question.

You are correct to note that if you are citing a work that is only available in Devanagari then an English-speaking reader would need a translation anyway, in order to read the cited paper. While this is true, you are still giving that reader less information about the paper than if you translate the bibliographic information into English, even if this translation is imperfect. You are also correct that the translated version might be annoying to a person who speaks the language of the paper at issue. While that is also true, such readers may be quite rare (depending on the journal, the subject matter, country, etc.). Writers usually try to ensure an adequate read for the majority of their readers, rather than imposing a language barrier that caters to a small minority. If you have reason to believe that a substantial number of readers of this paper would speak Devanagari then you might make the decision to include both citations --- e.g., use the original citation in Devanagari but then add a translation below.

  • 1
    Thanks for your answer, but I see one important point being missed. Some fields, like say, comparative literature, linguistics essentially deal with multiple languages. One just can't say that I will only access the literature available in English in the field of comparative literature. So while what you said is mostly true and agreeable for disciplines which have readers as described in your answer, some fields just require the knowledge (or the access) to the sources written in other languages where my question still remains.
    – Niranjan
    Apr 23 at 8:31
  • 1
    I've amended the answer to note these types of cases.
    – Ben
    Apr 23 at 8:37
  • 3
    Thanks for the amendment. I still don't agree with you. Having some idea, a little understanding, an approximate of the real pronunciation is nothing but the unwelcoming position to accept works in their original form. Also you seem to miss the difference between a translation and transliteration. If you look at the way I have written the citation, I still have the translation included, i.e., "A scientific grammar of Marathi" and I agree that to get an idea of what the work is about, this might be helpful. I have only excluded the transcription as it really feels a pointless activity.
    – Niranjan
    Apr 23 at 10:08
  • 13
    @Niranjan I wrote an article for the American Journal of Mathematics, and one of my sources is "Euclid, Elements". But I'll cite it as "Εὐκλείδης, Στοιχεῖα", and then call anyone who has difficulty with this 'unwelcoming'. Though to be fair, I didn't read the Greek original, I read a Chinese translation. So it would be more appropriate to cite it as "欧几里得, 幾何原本". Apr 24 at 7:23
  • 17
    "unwelcoming position to accept works in their original form." isn't it unwelcoming for the author not to consider the needs of the reader? The purpose of writing a paper is to communicate information, and the writer needs to consider their audience in order to facilitate that communication. In the case of the reference, it is to allow the reader to identifiy the original source and acquire it (or a translation). Apr 24 at 7:31

Being "fair" is not at issue. Being accessible to the audience of that publication is the issue.

As a professional technical writer and editor, my advice is: Do not make your writing needlessly complicated for the expected readers. If you make it less accessible for the readers, they will be less likely to read your publications. If they don't read them, they won't cite them, and your academic career will languish.

If I am reading a journal where the masthead is in Spanish and the publications are in Spanish, I expect the bibliographies to be in Spanish or at least in Latin alphabet. If I'm reading a journal in Korean, I expect Korean with bits of Latin alphabet.

In some journals I have seen the equivalent of this when the source was not in the journal's main language, sometimes with a link to an online version of the source or the publisher's listing:

Damle, Μ. K. 1970 A scientific grammar of Marathi. K. S. Arjunwadkar. Pune, India: Deshmukh & Company. (source in Marathi)

That is convenient because it lets me know what language I will encounter if/when I take the link. Although I don't read Marathi, I might decide to scavenge that work's bibliography to trace back an idea to its origin. I can search for a translation of the work and for others who have cited it (what did X say about it, versus Y?) i.n a language I can read.

With this form, on the other hand, even if I search for "a scientific grammar of Marathi" the first three pages of Google results have a whole lot of things about Marathi and Marathi Grammar, but no links to this book. I'm having a hard time finding your source.

दामले, मो. के 1970 शास्त्रीय मराठी व्याकरण [A scientific grammar of Marathi]. कृ. श्री. अर्जुनवाडकर. Pune, India: Deshmukh & Company.


I don't agree.

  1. French is completely different. If the article is in English, then all readers can more or less read and approximately pronounce the French words. If it is in a non Latin script, many of them can't. Nobody minds seeing non Latin scripts, but they do want to be able to read the information in some form, so transliterations or translations are necessary.
  2. Transliterating the title gives readers an exact or approximate version of the title in the original language. You do need to tell readers the basic information about the publication, including the author and title, in a way that they can understand, as far as possible. They might not be able to read the publication itself, but they need to have some understanding of what it actually is, and be able to refer to or cite it themselves (citing things without reading them is not good practice, but sometimes it is necessary).
  3. See answer to 2. For people who can read the original language and find transliterations annoying, you can include the information in the original language as well. And some languages do have almost universal transliteration schemes.
  • 3
    1. "If the article is in English, then all readers can more or less read and approximately pronounce the French words." Did you just forget me? I said I couldn't read and pronounce the word correctly and well you can't say that I don't know enough of English as I didn't get a French word right ;-) Your response to the first point sounds very assuming that everyone knows x whereas it is clearly not the case.
    – Niranjan
    Apr 23 at 9:50
  • 8
    1. I said "more or less read and approximately pronounce", not "pronounce correctly."
    – gib
    Apr 23 at 10:29
  • 7
    2. This was not my main point. My main point was that readers need to have some understanding of what the publication actually is. 3. Some transliteration schemes are almost universal, and some are not. What you are getting out of this exercise is being considerate to the readers, so they can have some idea what the publication is, including the title and author.
    – gib
    Apr 23 at 10:36
  • 3
    @Niranjan "2. "(citing things without reading them is not good practice, but sometimes it is necessary)" I don't and probably would never agree with this " does this mean that you would not acknowledge the original source of an idea because it appeared in a paper written in a language you did not understand? Consider the fairness of that practice if anglophone authors did not give due credit to academics outside that sphere. Apr 24 at 7:41
  • 2
    @Niranjan Re "3. How many transliteration schemes are we supposed to know?" - None. The point is that a transliteration provides some indication of pronunciation, so that we don't need to know about some other script. Imagine everyone using their own preferred script for their name when publishing (in English language venues). On the other hand, following your line of thought: How many scripts are we supposed to know? Apr 24 at 8:23

The example of Jean and John is not the same as transcribing a name into another alpahbet, is is a translation of the name, and might therefore make the author harder to find. You would not write a German Frank Müller as Frank Miller but rather as Frank Muller following your reasoning.

Using non-latin characters makes a paper less accessible and thus less inclusive. Think of blind people using screen readers. While it might please a minority of readers able to read that script, it will be burdensome to most others.

  • 1
    I would write Frank Müller only by my logic, I don't get it why you think that I would write Frank Muller. "Think of blind people using screen readers. While it might please a minority of readers able to read that script, it will be burdensome to most others." We are in the Unicode age. Everything is machine readable these days and by the way if at all something is not machine-readable at the moment; it will only become readable when people start using it! Discouraging people from using their scripts is highly authoritative and disrespectful (and by the way exclusive).
    – Niranjan
    Apr 23 at 14:52
  • The matter isn't of pleasing majority/minority. IMO it is a purposeless convention we are following. My point was if a source is supposed to be read (which is ultimately the motive behind citing) and it is written in some other language and script, the reader is obliged to have a translator and in such a case, it is really pointless and unjust to Latinize the names and titles is what I feel.
    – Niranjan
    Apr 23 at 14:59
  • 13
    @Niranjan Getting people to read the source is not the only motive behind citing
    – gib
    Apr 23 at 15:19
  • 5
    @Niranjan, to put it bluntly, the considerations of what bugs us as being "authoritative" or "exclusive" when we write are largerly irrelevant. We write for our readers, so please think whether you are doing any favor to your reader (or to the author you cite, so to speak) by doing what you are intending to do. Apr 24 at 8:31

You should give bibliographic information in a way that makes it easy to find in standard search engines. In math at least this means including a transliteration with Roman lettering. If you look up say papers written in Russian in zbMath or mathscinet they will have Roman transliterations that allow you to find all the other papers by the same author.

This has literally nothing to do with pronunciation, and in fact if the standard transliteration used by librarians is “wrong” from a pronunciation viewpoint you should use the standard one and not correct it.

People need to be able to easily locate the author and the author’s other works.


If the paper is written in the Latin alphabet, then that is an indication that the reader will be comfortable with the Latin alphabet, but may not be able to read text in the Devanagari script. The purpose of giving a reference is to specify a source and allow the reader to locate it, so giving the reference in the Latin script makes it easily available for the whole audience. Giving the Devanagari reference would make it more difficult for part of the audience that could not parse that script. Note that automatic translation (e.g. Google translate means that a Devanagari text may be understood to some extent by readers unfamiliar with Marathi)

If the paper were written in Devanagari script, it would make more sense not to transliterate the title as doing so could only obfuscate the source rather than identifying it for the reader.

The key point is that the purpose of academic writing is to communicate information, which requires respect for the needs of both the author and the reader and this may require compromise. Sometimes there is no perfect solution to a problem, and I think this is an example.

Of course if the source has an ISBN or a DOI or a URL etc. that makes things much easier.

The other consideration is making the reference accessible for automatic indexing systems (it would be better if the indexing systems were able to perform that transliteration). Also academics get recognition by their field via citation of their work. If I see a reference to an work and I can associate that citation with the author then that has enhanced the authors visibility. If the reference is in Devanagari script then I will not make that immediate association, which is to the author's disadvantage.

FWIW I would have no objection to my name being transliterated using Devanagari script in a paper written in that script.


Fundamentally the most important consideration is the journal's rules/expectations on citations and referencing.

If there is vaguity, the best approach would be to focus on providing transliterated options as Latin script is the most prominent script-form in academia and makes it easier to search for the citation. Whilst the consideration of respect and not making a butchering of someone's name in a transliteration is well and good, having just the non-Latin script name and title can make the reader's life hard if your paper is not in Marathi language and Devanagari script (i.e. the target audience is English speaking or another Latin script language).

For example, the source I am referencing is in a hypothetical language that uses Wingdings to show you how it feels to an audience that cannot read something that isn't transliterated into a script they understand:

☠︎♓︎❒︎♋︎■︎🙰♋︎■︎ (2022)📬︎ ❄︎♓︎⧫︎●︎♏︎ □︎♐︎ ♋︎♍︎♋︎♎︎♏︎❍︎♓︎♍︎ ◻︎♋︎◻︎♏︎❒︎

See how that means nothing to you and how it would complicate the reader's ability to figure out what article in your bibliography you're referring to?

Whereas a transliteration into the target audience's script (e.g. Latin) makes it easier to understand what you are referring to:

Niranjan (2022). Title of academic paper 

If you so wish (and the journal allows double listings in this situation), you could list the transliteration of articles as the primary way of mentioning it and then provide both the transliteration and also the original-language script version of it underneath in the bibliography:

Niranjan (2022). Title of academic paper 
[☠︎♓︎❒︎♋︎■︎🙰♋︎■︎ (2022)📬︎ ❄︎♓︎⧫︎●︎♏︎ □︎♐︎ ♋︎♍︎♋︎♎︎♏︎❍︎♓︎♍︎ ◻︎♋︎◻︎♏︎❒︎] 

Hope this helps!


Having just looked through a few East Asia-themed books at home, the norm indeed seems to be that the author's name is transliterated and then the title of the work is also transliterated. Sometimes, there is a translation of the title after the transliteration. I could have sworn I have seen works where the titles of sources were given in the original script, but that must have been in a library or so. I have found one Russian book that has cyrillic transliterations for Chinese and Japanese titles and no transliterations for titles in Latin letters.

Being somewhat able to read a few languages with non-latin characters, I tend to agree that transliterations of article titles are often of very questionable value. If you do not speak the language, they do not give a lot of information anyway. And if you speak the language somewhat, they make mapping the citation to the actual title of the work harder (similar to your point 3). I believe this is not so much of a problem in cyrillic scripts but it quite definitely is for Chinese.

On the other hand I feel quite strongly about transliterating the names of the authors (so readers can recognize them, as pointed out in other answers) and I am a big fan of translating the titles of such works.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .