6

I'm working on a written language where ligatures are important. Pretty much any letter could be joined to any other, and doing so could change the meaning of the word.

Too facilitate writing about this language digitally I'd like to transcribe it to Latin letters, but arbitrary ligatures are hard to write in a fluent manner. One idea that has occurred to me is to use some sort of diacritic to note that a letter should be joined to the next. Using a suitable editor (I use Emacs, enabling the TeX input method) this as is easy as pushing an extra key or two, and it can be done without interrupting the flow of writing. (Easy to write diacritics include t̄, ṯ, ṫ, ṭ, ẗ, t́, t̀, t̃ and t̂)

Question: Is there some method already in use to denote arbitrary ligatures, preferably denotable in Unicode?

Ideally, I would like to draw slurs above or below the letters, as is done in sheet music, but I haven't found a convenient method of writing this.

My current favorite is using a bar below a letter to denote that it should be tied to the next. It lo̱oks li̱ke ṯhis. I'm not sure though, as it might be more "logical" to underline both the conjoined letters, li̱ḵe ṯẖis.

The method needs to be able to denote ligatures with a̱ṟḇi̱ṯṟa̱ṟi̱ḻy̱̱ m̱a̱ṉy̱ ḻe̱ṯṯe̱ṟs̱.

A problem with this approach is that the under lines move around depending on the font used! If I copy some of my underlined text into the box where you can write an answer, the lines seem to move one letter to the right, and when posted they seem to be shifted more or less randomly. (You could try this to see for yourself!) Their position is nice and consistent when I write them in Emacs, but it would be nice to retain the option of exporting the text to some other format some day.

I've seen the method of using brackets, li[ke] [th]is, but I will be conjoining a lot of letters, and it sort of looks a bit clunky.

At this point I'm still open to diacritics, maybe some kind of brackets or perhaps even hyphens (lik-e t-his), or something completely different, I just thought knowing what others have done might help me to figure it out.

5 Answers 5

2

For completeness, I'd like to mention that Unicode does support putting slurs over an arbitrary number of characters…in theory. Characters U+1D177 "begin slur" and U+1D178 "end slur" (or U+1D175/6 "begin/end tie") are mainly meant for use in MusicXML and similar formats, but they're characters like any other. In theory, you should be able to use them in text, to put a slur over your letters.

Unfortunately, I don't know of any font that supports that usage, currently.

2

Consider adopting - or reversing - a convention that I have seen used in Catalan: “ll” is normally a digraph representing a single sound; there are words in which it should not be considered a digraph, but instead two separate occurrences of “l”. To indicate this, Catalan uses the interpunct (centred dot), “l·l”.

If, in your Latin alphabet transcriptions/transliterations, your ligatures are less common than isolate letters, use the interpunct to signal that the letters on either side are to be ligated (e.g., a·e is æ and ae is ae); if the reverse is true, use the interpunct to separate non-ligated letters (e.g., ae is æ and a·e is ae).

(The interpunct as a HTML entity is · or ·.)

2

Choose a notation you like. Choose a font where it looks good. Use those wherever possible.

You describe a font issue that’s causing problems for your chosen notation. This should be possible to work around at the font level!

Literally any text might be formatted in an undesirable way by some font, rendering engine, or combination thereof. It’s just hitting you hard in this case because of the combining characters. These often have bugs that may only be apparent on combinations of characters that the designers didn’t expect to be used (“nobody ever puts an ogonek on a letter t!”), or with a particular combination of font and rendering software.

For what it’s worth, my system[1] doesn’t have the “shifting” you report: the underlines in “a̱ṟḇi̱ṯṟa̱ṟi̱ḻy̱̱ m̱a̱ṉy̱ ḻe̱ṯṯe̱ṟs̱” appear in the same positions in your question and in this text box where I’m typing. Well, almost the same: the font in this box is monospaced, and also larger than that elsewhere, so I can clearly see spaces between each underline where I’m typing. In contrast, the proportional font used in question and answer text loses these spaces between narrow letters (like “itr” and “ril” in arbitrarily).

Hence my recommendation. Wherever it’s in your control, you can use the font that works for you. If you choose a freely available font, you can recommend the font to readers to try and cover cases where it’s not in your control. And if publishing your transcriptions on the web, which is an odd mix of “in your control” and “not”, you could use a font from Google Fonts or similar.


[1]: I know my setup isn’t especially typical, though. I’m on Fedora Linux 36, and my browser (currently Firefox 107) reports the fonts I’m seeing as Liberation Sans for question and answer text, and Liberation Mono for the text entry box.


Some further thoughts: Unicode offers a wealth of combining characters. Some may serve your purpose better than your current choice, which is U+0331 COMBINING MACRON BELOW.

  • For example, here’s U+0332 COMBINING LOW LINE instead:
    “a̲r̲b̲i̲t̲r̲a̲r̲i̲l̲y̲ m̲a̲n̲y̲ l̲e̲t̲t̲e̲r̲s̲”
    (This looks better to me, forming an almost unbroken underline. Almost: there’s a gap after the “m”, and the descender on each “y” pushes the line out of place.)
  • Some combining characters specifically tie two successive characters together. The most slur-like is U+035C COMBINING DOUBLE BREVE BELOW. This looks good for me in a monospaced font, but it turns into a bit of a mess with a proportional font:
    a͜r͜b͜i͜t͜r͜a͜r͜i͜l͜y m͜a͜n͜y l͜e͜t͜t͜e͜r͜s … a͜r͜b͜i͜t͜r͜a͜r͜i͜l͜y m͜a͜n͜y l͜e͜t͜t͜e͜r͜s
  • Another option is U+035F COMBINING DOUBLE MACRON BELOW:
    a͟r͟b͟i͟t͟r͟a͟r͟i͟l͟y m͟a͟n͟y l͟e͟t͟t͟e͟r͟s
    (This isn’t so good in the fonts my browser is using, but it looks much better in my text editor, which is using Bitstream Vera Sans Mono.)
  • There are also the Combining Half Marks, at least some of which are explicitly meant to be used to compose marks that span more than two letters. The trouble is finding a font that supports this properly! Another downside is that it can be fiddly to insert different characters for the start, middle, and end of your sequences.
    • This example uses U+FE2B to start each tie, U+FE2D for the middle, and U+FE2C for the end. To me, it looks okay, but not fantastic:
      a︫r︭b︭i︭t︭r︭a︭r︭i︭l︭y︬ m︫a︭n︭y︬ l︫e︭t︭t︭e︭r︭s︬
    • I get better results with the “continuous macron” characters used by Coptic (U+FE24, U+FE25, and U+FE26), but they’re an overline, not an underline:
      a︤r︦b︦i︦t︦r︦a︦r︦i︦l︦y︥ m︤a︦n︦y︥ l︤e︦t︦t︦e︦r︦s︥

Finally, markup may be more suitable than any character representation. For instance, a simple underline produces an unbroken connecting tie on most systems.

1

One idea (roughly borrowing from the practice of transcribing cuneiform) is make creative use of capital letters: Use lower case letters for single letters of your conscript and capital letters for the ligatures. When two ligatures follow each other without an intervening single letter, place a separator, e.g., the middle dot ·, between the groups of capital letters.

If you want to use slurs (I like that idea) have a look at MusicTeX (you will have to fiddle around a bit with it to apply the slurs to letters instead of musical notes).

1
  • Hm. Interesting. I can't believe I didn't think of capital letters! Thanks! I'm not sure it will turn out to best suit my purposes, but it's promising. MusicTeX is also a good idea, but it requires compilation. Capitals are easier and could be used by anyone. I'll wait a few days to see if there's more answers, but this might end up the accepted answer.
    – Edvin
    Commented Dec 13, 2022 at 12:34
1

In cuneiform studies, it's common to use "half brackets" to indicate a damaged section of the text.

a section of cuneiform text with half brackets

You might find these less disruptive than full brackets, in terms of the overall flow of your text. The usual Unicode characters for this are top left corner (U+231C) and top right corner (U+231D), but depending on the font they may introduce an annoying amount of whitespace into your line: a⌜eio⌝u. And unfortunately, while there are several options for the right half that take up less space ("end of stimme" a𝆨a, "combining left angle above" a̚a), I haven't found any good options for the left half. For the image above I used LaTeX to override the font's kerning.

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.