I have some Hebrew (right-to-left) text within an English (left-to-right) text as such:

enter image description here

The Hebrew text (right-to-left) by itself looks like this:

enter image description here

When the paper does not have enough width, the Hebrew text would wrap naturally as such:

enter image description here

What's the correct way to write the text when there is not enough space to finish the right-to-left quote in a single line?

Option A:

enter image description here

(note that we have to read the Hebrew text from bottom-up if this is the case)

Option B:

enter image description here

(aligning the start-of-line of the Hebrew text)

Option C:

enter image description here

(same as Option B but compacted left)

Option D:

enter image description here

Option E:

enter image description here

  • 2
    Oh, dear. A very difficult question. I don't whether some standard exists for this. I would probably pick D if I had to: that way it will at least be easy to read for speakers of Hebrew (the words will be in the right order locally), and there are no weird blank spaces. However, nobody says there has to be a solution to your problem: I think the best approach is to circumvent the issue entirely by putting the Hebrew on a separate line. I can't really think of a reason not to.
    – Cerberus
    Feb 22, 2013 at 19:59
  • 3
    @Alenanno No, it can't, due to the reasons of calligraphy. Unlike Japanese where the alignment does not impact the glyphs. If you practice writing Japanese or Chinese you may also notice that writing top-to-bottom is more comfortable, assuming the order of strokes. Feb 23, 2013 at 6:30
  • 2
    @bytebuster Yes it is indeed (I do practice them). :D
    – Alenanno
    Feb 23, 2013 at 9:59
  • 4
    Option D is what I am used to in Jewish publications, e.g. Artscroll, Koren, and other publishers. (If a scan of a typical page would be helpful let me know.) I'm talking about texts that are either predominantly English or about 50/50 (e.g. running inline translations); I don't know what happens for the reverse of this question, a Hebrew text with just a few English words. Feb 24, 2013 at 19:42
  • 1
    @SethJ, more natural to most readers.
    – Pacerier
    Feb 26, 2013 at 12:16

3 Answers 3


As a reader of Hebrew, I can tell you that the many times I've encountered such scenarios, it's been written with Option D, which is by far much easier to read. It is more natural to the reader, and this seems to be the norm in virtually all publications I've encountered.

In social media, and even on SE sites, it is very difficult, and in some cases impossible, to render it this way, and the result is something that appears to be gobbeldygook and takes a good deal of effort, sometimes even requiring contacting the writer, to decipher.


I say option D.

When in doubt, you can try what Web browsers do. They have touchy algorithms to handle this kind of cases, and not all Web browsers render them correctly. Firefox is quite good at it.

Richard Ishida has a good page of explanations where you can test your Web browsers. It's fun when it comes to numbers and parentheses.

Here are more detailed insights.

  • The problem is that they are looking for the easiest and most reasonable (not too much processing time) solution to the problem. What I want is the correct solution, irregardless of computer technology
    – Pacerier
    Feb 24, 2013 at 2:57
  • @Pacerier — Who, “they” ? Feb 24, 2013 at 10:11
  • Unicode personnel
    – Pacerier
    Feb 24, 2013 at 12:40

TL;DR: Option D is the answer, and Option E may be an answer with minor corrections.

I am affiliated with a company that develops a formatting engine, so let me tell how we do it in printing industry.

The decision point comes from an understanding of Unicode sequence, and difference of RTL (Right-to-Left) Unicode sequences to a plain text.

In a Unicode text, you will have the following sequence of character codes. E.g., The title says, then space and quotation mark, and then the string "פעילות הבינאום" in their natural, RTL, order:

  • פ
  • ע
  • י

and so on.

So, when a rendering engine performs typesetting, it takes those characters in their natural order:

  • It takes an English text;
  • a space;
  • a quotation mark;
  • (1) Note, there is no breakable space here
  • and the first Hebrew word, פעילות, starting from "פ";
  • then it takes another space symbol;
  • and then it finds out that the second Hebrew word, הבינאום, does not fit into the first printable line.
  • The rendering engine takes whatever fits and breaks the line at the nearest break location (a breakable space between the two Hebrew words).
  • Then it starts processing glyphs within the line: English text is rendered with LTR rules, while Hebrew text is rendered with RTL rules.

This distinction within the line (RTL vs. LTR) can be resolved in a different manner. Some formatting standards assume detection by text codepage; others, like XSL-FO, are being specifically instructed by writing-mode property.

So it will be naturally Option D.

You should also consider that RTL does not imply Bottom-to-Top. This is why Option A is not relevant.

Note that Option E is essentially the same as Option D, only having text alignment set to both, except that the quotation mark will be aligned together with the first word of the Hebrew text since there is no breakable space in (1).

I must admit that you have prepared your question very well, so I don't even need screenshots! :)

  • This may be how the computer do it, but is this the right answer? What I mean is, is this the way people would do it before the age of computers?
    – Pacerier
    Feb 24, 2013 at 3:01
  • @Pacerier I have no direct answer on this. I suppose that rules of writing combined texts have appeared long before computers and even typography. Say, there were some rules governing this. Call these rules X. At the moment of typography invented, typography should inherit X, instead of introducing another rule Y. Otherwise people would not accept typography at all. So chances are that X hasn't changed significantly, and therefore, today's computer rules are equal (or close to) the original rule set of X. Again, it's a pure speculation, so if someone suggests a better idea, I will accept it. Feb 24, 2013 at 3:27

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.