According to the Unicode charts, this is the psili or the smooth breathing mark used in the polytonic orthography of Greek, marking the absence of an /h/ sound in Ancient Greek for an initial vowel. It is also used to indicate the ejective or glottalized voice in Americanist orthography of several American languages, such as Hənqəminəm (see this link). So Unicode uses the same code points for Greek and Latin.

The combining turned comma above, U+0312, is not used in Greek, and I'm not sure there are orthographies where it is used together with U+0313. But I suspect they can be flipped versions of each other.

This has nothing to do with the caron, even though dcaron, lcaron, and tcaron have superficially apostrophe-like carons.

Is it correct to design this as a 180° flipped version of U+0312 «Combining Turned Comma Above» (which, I assume, is the one to use in a gcommaaccent)?

That depends on whether you intend it for Americanist use or Greek. The Americanist linguistic orthographies originated in the typewriter era, so the original form for that variant of U+0313 was an actual comma. The Greek spiritus lenis has a somewhat different shape, more tightly curved than a comma. If you have Arno Pro, take a look at it for an example. Arno Pro mostly doesn't have combining diacritics, but you can look at U+1FBF (or U+1FBD), which is the spacing form of Greek U+0313. The spacing turned comma in Arno (U+02BB) is quite different, although in combination with g (U+0123) the difference is less.

Thanks! I'm surprised actually to hear this is used in Greek – I sort-of assumed it was intended for some Latin Extended context, probably because I was looking at it next to U+0312. Curious.

"the original form for that variant of U+0313 was an actual comma. The Greek spiritus lenis has a somewhat different shape, more tightly curved than a comma."
Ahh, is that by any chance the «accent»/«breathing» difference that Nick Shinn shows here? http://www.typophile.com/node/34728#comment-210877

I don't have Arno Pro unfortunately. But it occurred to me to check how different fonts do it via fileformat.info. Indeed it seems common usage to make U+0313 look much like a comma.

Yes, sort of. I mentioned Arno just because they put a lot of good work into the polytonic Greek for it, but lots of other fonts are relevant too. Just for reference, here are the shapes in Arno, but Nick's examples work very well also.

In Greek, U+0313 is the combining form of U+1FBF, and they should look the same. They are both breathing marks, and both should look like Nick's "breathing" form.

The problem is that Unicode has combined two similar but slightly different forms in U+0313. However, you probably aren't making a polytonic Greek character set. In a font that has extended Latin support but no polytonic Greek, a comma shape (Nick's "accent" form) is appropriate for U+0313. And actually even in a font with both extended Latin and polytonic Greek, if you have precomposed glyphs for all the polytonic Greek letters with breathing marks, then the U+0313 grapheme can safely be comma-like, since it will only be seen in Latin contexts.

Of course, the number of people who use polytonic Greek is probably several orders of magnitude larger than the number of people who use Salishan languages, not that either is of much commercial significance :)

Just one remark. Curiously enough, Unicode defines 77 precomposed Greek characters with psili from U+1F00 to U+1F30, for instance U+1F00, U+1F02, U+1F04. You can see them in the following Unicode chart: http://www.unicode.org/charts/PDF/U1F00.pdf. I downloaded the xml file containing Archimedes Method in polytonic Greek. The character U+0313 does not occur even once in the file. There is however plenty of precomposed unicode tharacters with the psili.

The character U+0313 does not occur even once in the file. There is however plenty of precomposed unicode tharacters with the psili.

If you take a look at the Unicode code chart, you'll see that these precomposed characters decompose to sequences including U+0313. In theory (for the sake of machine text processing), the correct way to encode Greek in Unicode is to use the base character plus combining modifiers, but in practice people tend to use the precomposed forms, because these are much more widely supported by current software. The precomposed glyphs in the extended Greek section are a legacy of older standards that used different principles being folded into Unicode, sort of like the presentation forms in extended Arabic.

The precomposed forms contain all the glyphs that you need in normal everyday use, but for prosodic marking or certain situations involving transliteration from epichoric alphabets, you still need combining diacritics.

Polytonic Greek and Americanist orthography are unlikely to overlap in usage, but it's still a bit of a shame that Unicode decided to unify the combining psili with the combining comma above, considering the different design traditions for these two uses. But then there are other cases of unfortunate unifications in Unicode.

I agree with David (archaica)'s recommendation about the design of the comma above. He has explained it nicely.

Is the combining comma above used frequently in American orthographies? Many Salishan languages seem to use the modifier letter apostrophe to mark ejectives after the consonant, following IPA, instead of putting the comma above the consonant.

"In a font that has extended Latin support but no polytonic Greek, a comma shape (Nick's "accent" form) is appropriate for U+0313."
That's sort of the gist of what I wanted to find out; but I'm a bit reluctant to say that, as I'm quite fascinated by the wealth of information and the insights that have surfaced in this thread, which I wouldn't want to shut up. Thank you guys.

Good old U+0313. The combining comma is preferred by some communities and/or linguists, but other communities and/or linguists prefer the modifier apostrophe instead. (Linguists have often been the main arbiters of orthographic decisions, but this is changing nowadays.) Although for a while now the variation has been due to technical capabilities among computers and typesetting systems, it’s largely an aesthetic issue at this point.

When reproducing older texts, such as those gathered by linguists in the early 20th century, using the combining comma is obligatory if one aims for precise reproduction. And precise reproduction of old materials is becoming very important with the spread of digital language archives, endangered language corpora, and other language documentation efforts.

BTW, the combining comma was used for quite a few other language families besides Salishan – and its neighbors Wakashan, Chimakuan, Chinookan, etc. – since nearly all languages in western North America have one or two ejectives. I’ve also seen it used for Mesoamerican and South American languages, Siberian languages, and Caucasian languages. Basically anywhere an American linguist might have encountered ejectives, there you’d find the combining comma pop up.

In linguistic transcription it’s much less common in new work as people move to IPA, but transcription traditions in some circles can be extremely hard to change. In addition, historical linguists prefer to use older symbols rather than invent new ones due to the structure of their science, and so even people otherwise using IPA – e.g. me – will maintain crufty old Americanist transcriptions in certain situations.

I personally dislike the combining comma, but I think my dislike stems largely from its poor implementation in many fonts. I don’t mind it too much when reading works set by the US Government Printing Office back before WWII, for example. (The USGPO printed for the Smithsonian, among a few other sources of linguistic research back then. Someone needs to do a typographic study of their practices and products over the years, which are a standing testament to the versatility of Scotch Modern.)

As for its relationship to the caron as used in dcaron, lcaron, tcaron, I don’t think the combining comma should be considered related at all. The comma-accented caron is in my experience always an apostrophe shape, rather than the slightly smooshed appearance of many true combining commas. The two diacritics are in any case very different in their functions. You might also consider h-caron, i.e. U+0215 ȟ, which lacks the comma-accented form despite having an ascender because it lacks the weight of typographic tradition. It’s used for IPA /χ/, a voiceless uvular fricative, in some American languages.

James, thank you for your very informative post. In the Lushootseed font for The Tulalip Tribes, we used uni0313. The shape is a crescent, not a comma, being a traditional form in Salish art whereas the typographic comma shape (ball-and-tail) is quite foreign.
Juliet