Not signed in (Sign In)

Vanilla 1.1.9 is a product of Lussumo. More Information: Documentation, Community Support.

    I notice "Kahler" appearing not only in postings, but in tags. "Kaehler" is a perfectly acceptable substitute for "Kähler" when one cannot type an umlaut --- "Kaehler" and "Kähler" are essentially the same spelling. If "Kahler" were the right spelling, it would be pronounced differently. Could spelling it correctly in tags be the beginning of making people aware of this?

    Why, do you know mathematicians who pronounce it (approximately) like 'caller'? I've never heard anyone pronounce it like that.


    There is at least one good argument against the Kaehler spelling, as far as MO is concerned.

    When you use Google to search for " kahler" you get results for both "Kahler" and "Kähler" but you don't get anything with the "Kaehler" spelling. Similarly, when you search for " kaehler" you only get results for "Kaehler," but no results for "Kahler" nor "Kähler". In other words, "Kahler" is usually better than "Kaehler" for searching purposes.

    (Sadly, the primitive MO search tool treats Kahler, Kähler, and Kaehler as three different names...)

    @François: How sure of that are you? I've done Google searches where I entered a word with "ae" and I got results involving words with "ä" where I had written "ae". And that is exactly what I had intended. (Similarly with "ue" and "oe".)

    @Todd: I've never heard anyone pronounce it as if it's spelled "Kahler".

    On Google (from UK), I find the following results when I seach for XXX where XXX is kähler, kaehler and kahler:

    • Kähler: about 386 results
    • Kaehler: about 170 results
    • Kahler: about 876 results (but searching for kahler -kähler yields about 173 results)

    I agree with the implicit request at the beginning. Let the tag include the umlaut. Possible?

    At, it says "Type to find tags:". So in the search box, I entered "Kähler" and got "Kahler-differentials", and then I entered "Kaehler" and got nothing.

    One of the correct spellings redirects to the incorrect spelling (which is in effect the official spelling in mathoverflow tags).

    The other correct spelling does not.

    I just reran José's experiment from Canada and I got different numbers:

    • kahler - about 873 results
    • kaehler - about 173 results
    • kähler - about 908 results

    After inspection, it appears that Google Canada does recognize Kaehler as an alternate spelling of Kähler, but not the other way around.

    Then I used VPN to repeat the experiment from the US and I got results that match José's UK results:

    • kahler - about 882 results
    • kaehler - about 172 results
    • kähler - about 387 results
    I thought that there was an unwritten convention that "it is acceptable to drop accents". I am certainly not saying this is a good thing, and I might even wince when I see etale cohomology rather than \'etale cohomology -- but that's the way of the world and given that there are hundreds of millions of US citizens out there all using keyboards that have no apparent \' or \" buttons (and also I guess that a generic person may well not know that \"a and ae are the same) I can see how this has happened and also how hard it would be to change...
    • CommentAuthoran_mo_user
    • CommentTimeJul 5th 2011 edited

    Two short points:

    1. Since Kähler is a German name, the correct spelling (in a German context) in case no diacritical marks are available is 'Kaehler'. (The German pronuncaiations of 'ä' and 'a' are very different, the 'ä' is close in pronunciation to a German 'e' and not at all an 'a'.) More generally, in a German text poduced under circumstances were no diacritical marks are available all the umlauts are transcribed by ajoining an 'e'. So for example 'für' (meaning 'for') becomes 'fuer' and definitely not 'fur' (which looks completely wrong in a German context); they exist for the vowels a,o,u. [Rarely there also vowels in particular 'i' and 'e' with two dots which are not umlauts, but rather the diacritical mark is 'imported' from a foreign language, mainly French, and in this case this apending of an 'e' in the absence of the diactritical mark is not performed; but rather the diacritical mark omitted.]

    2. Having said 1. I still think that 'Kahler' might be the best choice. [Sort of following MathSciNet that while having the diacriticals marks available essentially ignores it for searches. Searching 'Kahler' will yield 'Kähler' and (of course) 'Kahler' but 'Kaehler' won't yield 'Kähler'. Also searching 'Kähler' yields 'Kähler' and 'Kahler'.] The reason for this is that while I happen to know what is going on in German, I am aware that I am totally clueless regarding the analog problem in almost all of the many other languages where similar issues exist, and the simplest fiable rule applicable over all languages to me seems 'ignore all diacritical marks for tags'.

    ADDED: Or, in short: what Kevin Buzzard said (which was not arround when I started to write this).

    A convention that favors dropping "accents" seems to regard that as a necessity, but in this case there's a standard accentless alternative: Kaehler.
    • CommentAuthorWill Jagy
    • CommentTimeJul 5th 2011
    What is correct for "a priori" ? I have a new question with that and can do accents in Latex, plus I would prefer to reduce the grammatical confusion of the a with the word "a". Note that this is a standard phrase in PDE and that Gilbarg and Trudinger talk about "Apriori bounds"
    "A priori" is just from the Latin and there's no accent on it, right? Or did I misunderstand what you're asking?

    Michael Hardy, I think this is an interesting point, and one in favor of the Kaehler you suggest. A problem here is as Kevin Buzzard (an me, more implictly) said is that one has to know this standard.

    My worry and/or question (and while this being a bit off-topic I would be independently of the outcome interested to learn this in case somebody can comment) is that what we are discussing here is one (standard) convention. But, I suspect there are similar things for other languages. I somehow believe to know that in French one repalces (or at least might do so) accents by apostrophes and alike in case one does not have the accented letter available, so it would be e'tale cohomology and for the cédille one can use a comma so François would be Franc,ois. [Actually I am not really sure how much this is done nowadays, and this might be more a type-writer-age thing; yet Serre in a letter to Grothendieck asked him to use this c, construct, in combination with returning the carriage a bit, to get an almost correct ç because just c instead of ç was no good.]

    The question that arises for me is where does one stop if one starts doing this. For example, what about Čech cohomology? Is there some standard way to transcribe Č ? I don't know. I could do it in LaTeX but this is still something else.

    A complicated solution could be to requirer LaTeX-code (or some pseudo-LaTeX), but this seems overly complex and might really be bad for searching

    • CommentAuthorWill Jagy
    • CommentTimeJul 5th 2011
    Thanks, Kevin. That's exactly the question. I really thought there was an accent. At least, I am pretty sure I have seen one at times. So, putting you in the position of a character in a current U.S. television commercial (for Metamucil, a fiber supplement), "And those people are what I like to call `wrong'".

    Will, I suspect the French are the culprits for your confusion. A few common Latin phrases have been Frenchified (which is how Google translates franciser). Among them are a priori, a posteriori, and a fortiori which are commonly spelled "à priori", "à postériori", and "à fortiori". The French generally prefer the Frenchified spelling, but both are acceptable so long as the first is italicized and the second is not.

    • CommentAuthorWill Jagy
    • CommentTimeJul 5th 2011
    Thanks, Francois. I needed the phrase for my question, although I know Pete does not see the overall question in the same manner. There has got to be a better alternative to Frenchified, there is "anglicised" for English.
    • CommentAuthorgrp
    • CommentTimeJul 5th 2011
    I imagine there are those opposed to the position (held by some Frenchmen and Francophiles) of keeping the French language pure or by adopting and transforming foreign phrases into something that looks like native French; such opponents might describe such altered phrases derisively as "French-fried". (Of course, I would never do that. Oh no, not ever.) To which the appropriate response would be (forgive my poor attempt) "Non, c'est pomme frites."

    Gerhard "Crawling Back Into My Lair" Paseman, 2011.07.05
    Sometimes one italicizes things like "a priori" because they're foreign. That might prevent some people from confusing "''a''" with "a".

    A certain German pop-singer has written that when she was eight years old, being accustomed to seeing her father use a typewriter, she decided she wanted to learn to type, and got the machine out of its dust cover and started working on it all alone. She couldn't find the "ö" key and concluded that its inclusion had been neglected, and started loudly expressing her frustration and anger, thereby getting the attention of her mother in the next room. Her mother then pointed out where the "ö" was. (But this paragraph is irrelevant, so don't read it.)
    • CommentAuthorWill Jagy
    • CommentTimeJul 5th 2011
    I like the idea of italics for the whole phrase.

    You've got to explain the joke about the pop singer.
    There's not much to explain. Maybe I found it amusing because it's true. I couldn't help thinking of it when I read some of the comments above.
    One of the moderators has told me that this whole matter is one of those things where everyone involved is at the mercy of the software, which no one can change.

    ... until we move to SE 2.0, should that happen ...

    The correct spelling in French is "a priori", without any diacritic. I have never seen a dictionary or grammar book or heard any francophone claim otherwise ,
    You might read a French text with an inappropriate accent, but then you may also read French texts with any number of other spelling mistakes.

    Georges, you are right that the unaccented spelling is the only truly correct one. However, the spelling "à priori" was in the Dictionnaire de l'Académie from the 6th edition until it was suppressed in the 9th edition. The accented spelling also appears in Littré, but with a cautionary remark saying that this is wrong. See and references there.


    Thanks for your interesting and erudite comment, François. I think that indeed the French frenchified more in the past. They didn't even stop at proper names: Jérôme Cardan for the mathematician Girolamo Cardano is a notorious example.

    @ an_mo_user: "The question that arises for me is where does one stop if one starts doing this."
    So why stop? Just give "..higher priority to ensuring utility for the future than to preserving past antiquities." (Joseph D. Becker from Xerox in 1988.)

    "A complicated solution could be to requirer LaTeX-code (or some pseudo-LaTeX), but this seems overly complex.."
    LaTeX clearly is no solution! And obviously there is no simple solution. But there is a solution.
    (1) Use (ISO) transliteration instead of transcription and (2) for coding use Unicode.

    "For example, what about Čech cohomology?" "... might really be bad for searching"
    I entered Čech into Google and got 64.500.000 results. Seems to work well.
    (See for instance also: )
    • CommentAuthorquid
    • CommentTimeJul 19th 2011 edited

    Bruce Arnold, thank you for the information (btw, I am an_mo_user renamed). It is however not quite clear to me what your actual suggestion is, and the fact that you quote me in part out of context does not help. Though I found some of the information you provide interesting.

    To recall, the suggestion of Michael Hardy was to use 'ae' instead of 'a' for (a German) 'ä'. Using 'ä' itself is a non-option as this is not supported in tags. Likewise, 'Č' or 'ç'.

    In more detail: I just tried to create the tags Kähler, Čech, François on the test site; and while nothing was actually creteated (as I am 'new' on the test site) the system informed my that it could not create the tags kahler, cech, francois; so that it seems a given that it would have created (if I were not 'new') kahler, cech, francois and not Kähler, Čech, François. And, trying your Čech it tells me the tags 268 and ech cannot be created. So, while in principle it would be nice to have more characters available for tags, it is simply not so.

    Thus, we are limited to characters without diacritical marks. Unfortunately, the ISO transliteration you mentioned contains characters with diacritical marks, and is thus a non-opition, too. Yet, the GOST one, which I found on the site you link to, in fact does not. So using GOST transliteration would indeed be an option for Cyrilic. Now, as the next step we only need something like this for other alphabets, too. (Whether persuing this next step is useful in view of the to be expected likelihood of success of the one afterwards, i.e. convcing people to use this, I will let others decide.)

    And finally, my "...might be really bad for searching" clearly referred to the useage of latex-code or some pseudo version of it (so \check{C}ech or check{C}ech something like this); why you try to refute this by searching for Čech is unclear to me.

    Quid, I tried more to comment on the problem in general. I also just had read Qiaochu Yuan comment yesterday (70655) why he wrote "Pontrjagin" instead of "Pontryagin": "Both spellings seem to be pretty widely used. My impression is that Pontrjagin is a slightly more faithful romanization. "Pontryagin" invites a pronounciation in which the "y" and the "a" are pronounced separately, but in Cyrillic "ja" is one character and, as I understand it, one syllable." This points in the wrong direction. I agree with KConrad how answered: "In any case, languages aren't required to make foreign names or places sound like they do in the native language."

    The problem of tags and the limited flexibility of the system as it is at present is a much more special problem. "And finally, my "...might be really bad for searching" clearly referred to the useage of latex-code or some pseudo version of it (so \check{C}ech or check{C}ech something like this); why you try to refute this by searching for Čech is unclear to me." In the contrary: I did not want to refute your claim. Rather I tried to say that using Unicode would not lead to similar problems, with Google at least.
    • CommentAuthorquid
    • CommentTimeJul 20th 2011

    Bruce Arnold, thank you for the response. I only commented on the question at hand, the tags, and thus (mis)understood your first response to me differently; sorry for the confusion, good that this is clarified. In general, I agree that Unicode is a good thing.

    Regarding the (new) subject of transliteration of Понтрягин. I am not the right person to discuss questions on Cyrilic, and to judge whether the comment you quote points in the right or wrong direction.

    But, it is my understanding that Pontryagin and Pontrjagin both follow standards for transliteration (BSI and GOST (also DIN), resp); yet neither the one you suggest (ISO); this would give Pontrâgin. And, it seems to me (though I could well be wrong) that ISO actually tries to accomplish what Qiaochu said, namely, stress that there is rather one than two 'thing(s)' between 'r' and 'g' (while this is harder to infer from 'ya'). So, while I can see arguments for using BSI (the site being in English and this looking nicer and being more wide spread than the i^a of ALA), I can also see arguments for GOST (along the lines of Qiaochu's comment and avoiding the non-standard letter â of ISO) or also of course for ISO.
    So, in some sense, though this might be wrong, it seems to me that to use GOST is rather closer in spirit to your suggestion of using ISO, than using BSI would be, which leaves me again a bit cunfused.

    quid, you certainly know much more about these things than me.

    With regard to ISO, DIN, GOST, BSI: Ultimately we can escape the difficulties only if we agree on /one/ standard. Perhaps it is easier for people from different cultures to agree on ISO even if they see weak points. With Unicode this is already achieved.

    With 'wrong direction' I meant stressing the point of pronunciation, better said: of a transcription which is oriented at phonetics. All what is really important here is the basic quality of a transliteration: to be a map which allows the unique reconstruction of the original item from the transliterated.

    As KConrad said: ".. languages aren't required to make foreign names or places sound like they do in the native language." (And if you ever heard French people talk you know that they make ample use of their right to do so :)
    • CommentAuthorquid
    • CommentTimeJul 21st 2011

    Bruce Arnold, thanks for your first sentence and the explanation. But, in fact I learned quite a bit from this discussion; so, I thank you! Ultimately, for this site, in my opinion one should not worry too much about things like this. My reasoning is that while some people (e.g., me) find questions/discussions like this genuinley interesting, I can imagine that others find them tangential (in this context), in particular as this is a medium rather with quick turnover, and not like writing a book. So that trying to enforce something could annoy some.

    Regarding your last sentence: yes! (though I would not single out the French, as I also heard plenty non-French trying to say French names/things).

    In any case, thank you for the interesting discussion!

    • CommentAuthorsisn
    • CommentTimeJul 21st 2011
    Well Kahler is certainly very confusing for people who speak german (I am a native speaker), and I would have never thought about searching Kahler when I was looking for Kähler. I would have searched Kaehler, so I would second Michael Hardys request.
    • CommentAuthorquid
    • CommentTimeJul 21st 2011 edited

    sisn, so you do not use MathSciNet?

    Added: sorry, on second thought the above is perhaps too cryptic. First, it is not so uncommon that for searches diacritcs are ignored. Second, here this goes both ways: if you search among the tags for Kähler, you do get kahler-differentials. You would only be in trouble if you searched for Kaehler. It seems to me that the number people searching for Kahler plus people searching for Kähler will be significantly larger than people searching for Kaehler.

    > You would only be in trouble
    > if you searched for Kaehler.

    At I searched for kaehler and among the first 20 items, 17 say "Kähler" and three say "Kaehler". At the first 21 entries say "Kaehler" and the 22nd is the (English) Wikipedia article titled "Kähler manifold".

    @Michael: I'm assuming you are in Germany right now... can you repeat the same experiment that José and I did from there?