Modifier Letter Rhotic Hook

Paul Miller

There is a letter in the unicode standard 'Modifier Letter Rhotic Hook' at $02DE.  I was wondering what it is used for and who uses it ?

Which languages (if any) use this character and is it worth including in a font ?

I have looked on the internet but apart from it appearing in the unicode standard there seems to be very little information about it.

I suppose the main question is 'Is this character in wide useage ?'

George Thomas

These pages may be what you are looking for:

https://en.wikipedia.org/wiki/Rhotacism_(sound_change)

https://en.wikipedia.org/wiki/Rhoticity_in_English

::edit:: Ignore those links. I just realized they apply to consonants.

James Puckett

Only eight characters use the hook, they’re rarely used, and they’re in Unicode. So it seems like a character that’s probably never going to be used.

Cory Maylett

It's a mark used in the International Phonetic Alphabet (IPA) to denote an r-colored or rhotacized vowel used in some regional accents.

For example, in most American English, a hard r is used in words like, well, hard. One exception being the accent typically associated with Boston. Most British accents tend to drop the r in these kinds of words (again, with regional exceptions).

The rhotic hook is added to the right side of the preceding vowel to indicate that, in a particular regional accent, the r that follows it is pronounced.

As for including it in a font, probably not, unless you also include all the other IPA diacritics.

James Puckett

Only eight characters use the hook, they’re rarely used, and they’re in Unicode. So it seems like a character that’s probably almost never going to be used.

Paul Miller

Thank you all for the information.  I will leave it out.

I just wondered because 'Times New Roman' has this character as a modifier character which can be added to 'A' to 'Z' and 'a' to 'z' plus some accented characters.  It is in the open type features as a modifier character which actually extends the advance width of the character to which it is applied, unlike a diacritic which doesn't affect the advance width.

I thought there must be some hidden importance which I had missed.

Peter Baker

I read somewhere not long ago (but, sorry, I can't find it again now) that, since the only combinations of letter + rhotic hook that ever occur are already in Unicode, 02de is only useful if you need to discuss the rhotic hook.

Reading that gave me great pleasure.

Jens Kutilek

What are the other Unicode codepoints of all the vowels with rhotic hook? I know only two: ɝ (U+025D) and ɚ (U+025A).

Peter Baker

There's just those two. In a book called The Unicode Cookbook for Linguists, of which I can see a few pages via Google Books, the authors seem to consider this a problem, and they allude to "other combinations of vowels with rhotic hooks" which need to be made with 02DE. An official IPA chart shows as an example a with rhotic hook. Isn't it a problem, though, that 02DE is not meant to be a combining mark? What would be the best way to handle this?

Jens Kutilek

@James Puckett was mentioning eight characters that can combine with the rhotic hook.

 I did some more research and IPA seems to define 28 vowel characters. In theory, each should be able to combine with the rhotic hook.

Most of these could be handled with a mark positioning feature to combine nicely, but the ones marked here in red need special attention (these include the two characters that are defined in Unicode). Those combinations could be replaced by a precomposed glyph when they occur.

André G. Isaak

I’m pretty sure that it would be rather difficult to produce a rhoticized vowel which isn't in the neutral (mid-central) position (since that’s essentially where r is articulated) as a monophthong. I suspect that’s why IPA only encodes [ɚ] and [ɝ]. Anything else would likely be written as a diphthong (e.g. [iɚ̯] — essentially how a non-rhotic dialect of English might pronounce the vowel in 'fear').

Peter Baker

I've just looked at a couple of fonts that I would expect to do this intelligently. Doulos SIL makes 02DE a spacing modifier letter with a negative left sidebearing so that it overlaps with the preceding character: then the next character along respects the width of 02DE so there are no collisions. Doulos also includes a couple of anchors so that the positioning of the mark can be fine tuned:

Brill does essentially the same thing, but without the anchors—perhaps because some apps appear to ignore the anchors on this glyph?

Anyway, it looks (contra my earlier post) as if the 02DE really is useful.

Paul Miller

<sigh>  maybe it would be best to include it if I want my font to be complete.  However I don't think the approach adopted by 'Times New Roman' is correct.  I will include the anchors so it can be attached to the characters in the graphic in Jens Kutilek's post.  Perhaps I will take a look at Doulos and Brill just to see how they work.

Jens Kutilek

André G. Isaak said:

I’m pretty sure that it would be rather difficult to produce a rhoticized vowel which isn't in the neutral (mid-central) position (since that’s essentially where r is articulated) as a monophthong. I suspect that’s why IPA only encodes [ɚ] and [ɝ]. Anything else would likely be written as a diphthong (e.g. [iɚ̯] — essentially how a non-rhotic dialect of English might pronounce the vowel in 'fear').

Wikipedia says:

The r-colored vowels of General American can be written with "vowel-r" digraphs:

• [ɚ]: hearse, assert, mirth (stressed, conventionally written [ɝ]); standard, dinner, Lincolnshire (unstressed)

• [ɑ˞]: start, car

• [ɔ˞]: north, war

So there are at least a few more common ones in addition to the two Unicode characters.

Christian Thalmann

Mandarin apparently has tons of rhotacized vowels:

Wikipedia Link

Peter Baker

Christian Thalmann said:

Mandarin apparently has tons of rhotacized vowels:

Wikipedia Link

In the Wikipedia Erhua article Christian links to, 02DE rhotic hook occurs only via precomposed 025A (ɚ) and after u and 028A (ʊ). If the article can be trusted (always a question with Wikipedia), the situation could definitely be worse.

But I'm having trouble finding out what the phonetic difference is between the rhotic hook and 02B5 "MODIFIER LETTER SMALL TURNED R WITH HOOK". The table in the article Christian links to uses both.

(BTW: As a senior citizen with senior eyes, I'd like to register a mild protest against fonts that render these modifier letters very small and thin. I'm getting a crick in my neck leaning close to the screen to figure out what's going on!)

John Savard

Paul Miller said:

<sigh>  maybe it would be best to include it if I want my font to be complete.

That depends. It only makes sense to do so if you want your font to provide full IPA support. If you do, it is quite possible that the modifier letter rhotic hook is not the only issue you will face.

So before expending effort on resolving this issue, it would be advisable to check into how many other issues you will face, and their difficulty.

Simon Cozens

Peter Baker said:

There's just those two. In a book called The Unicode Cookbook for Linguists, of which I can see a few pages via Google Books

FWIW, I had the pleasure of proof-reading this book. The main that thrust of the book is the final two chapters which recommend a way for linguists to store information about orthography profiles for languages they're working on, and the first six chapters feel like they're working up to that, but it's still worth a read. As a langscipress book it's open source and the whole thing is available on github at https://github.com/unicode-cookbook/cookbook

Peter Baker

This is great! I've grabbed a copy, and very glad to have it. Three cheers for free scholarship!

Denis Moyogo Jacquerye

To summarize, the rhotic hook is used in the International Phonetic Alphabet and can be attached to any r-colored vowel (like in bird, car, dinner, etc. transcribed [ˈbɜ˞d], [ˈkɑ˞], [ˈdɪnə˞], etc. for American English pronunciation). It should not be confused with the retroflex hook, which shared the same function in the past, but is now only used with retroflex consonants. The rhotic hook is a generalization of the hook on early r-colored vowel symbols.

All the other symbols mentioned are, from my point of view, historical only when talking about r-colored vowels. However ɚ and ɝ (single characters) are still widely used instead of ə˞ and ɜ˞ (ə and ɜ combined with ˞).

Representation of R-colored or rhotic vowels was a hot topic in IPA discussions for a good part of the 20th century. Initially the IPA didn’t differentiate them from vowels followed by rhotic consonants (the various r-like consonant) or from the syllabic r-like consonant. Eventually linguists wanted to differentiate the r-colored vowels, they just couldn’t agree how.

For context, the symbols for retroflex consonant with the retroflex hook ʈ, ɖ, ɳ, ɭ, ɽ, ʂ, ʐ were adopted in the IPA in 1927 borrowed from Scandinavian phonetic transcription systems.

In 1935 John Samuel Kenyon used the symbols ɚ and ᶔ for the two American English unstressed and stressed rhotic vowels in American pronunciation: a textbook of phonetics for students of English. It seems the initial design of Kenyon’s symbols had a hook attached where the stroke of the vowel ends.

In 1939, the second symbol can be seen with the modern ɝ shape in the article “A Petition” in American Speech. Here the hook is the same as in the first symbol.

In 1947, the IPA finally adopted various symbols for rhotic vowels, and in 1951 it also adopted the first of Keynon’s symbols. The 1951 IPA chart shows five possible ways to represent the r-colored central vowel [ə]  central vowel: əɹ or əʴ or ɹ or ᶕ or ɚ.

The use of the retroflex hook for r-coloring in ᶕ but also ᶏ ᶐ ᶒ ᶓ ᶔ ᶖ ᶗ ᶙ was withdrawn from IPA in 1976, on the basis that it was practically not used for rhotic vowels. That means that the retroflex consonant symbols were and are still valid IPA.

In 1989, after a big clean up of the IPA symbols, the rhotic hook was adopted as the sole symbol for rhotic vowels. The other symbols are sometimes still in use by linguists who didn’t get the memo.

Unicode has a modifier letter rhotic hook <02DE>  ˞ and a combining retroflex hook <0322> ◌̢, but also the letters <025A> ɚ and <025D> ɝ.

However these are not canonically equivalent to <0259,02DE> ə˞ and <025C,02DE> ɜ˞ for unclear reasons. If they were equivalent, text processing could treat them as semantically the same.

My guess is <025A> ɚ and <025D> ɝ can better represent Kenyon’s symbols while 02DE can be used as the 1989 generalized rhotic hook symbol. But in practice they are used interchangeably with the sequences.

The historical symbols ᶏ ᶐ ᶒ ᶕ ᶓ ᶔ ᶖ ᶗ ᶙ where added to Unicode and are not canonically equivalent to their vowel symbol followed by the combining retroflex hook, for example <1D95> ᶕ is not equivalent to <0259,0322>, because the retroflex hook position is ambiguous.

As for the design of the rhotic hook, it should indeed connect with the previous symbol, with negative left side bearing or other positioning behaviour.

Peter Baker

Thank you, Denis. This is an extremely useful explanation.

Adam Jagosz

André G. Isaak said:

I’m pretty sure that it would be rather difficult to produce a rhoticized vowel which isn't in the neutral (mid-central) position (since that’s essentially where r is articulated) as a monophthong. I suspect that’s why IPA only encodes [ɚ] and [ɝ]. Anything else would likely be written as a diphthong (e.g. [iɚ̯] — essentially how a non-rhotic dialect of English might pronounce the vowel in 'fear').

I think what is difficult is subjective. The vowel quality for sure changes and maybe that's what might be in some far-fetched cases called rhotacism? For instance in

Denis Moyogo Jacquerye said:

[ˈkɑ˞],

I would say in some dialects the upcoming [ɹ] may be sensed and expected and heard quite early into the [ɑ] (through this subtle yet present coloring), but I would still transcribe it as [ˈkɑ˞ɹ] (or maybe there are dialects that actually only perform the r letter as the vowel coloring, [ˈkɑ˞]?). Conversely, it always made me wonder how linguists insist on denoting the nasality of vowels preceding [n], where in reality I didn't hear much of an [n] before the actual [n] started to ring.

Christian Thalmann

Conversely, it always made me wonder how linguists insist on denoting the nasality of vowels preceding [n], where in reality I didn't hear much of an [n] before the actual [n] started to ring.

Compare the words «ban» and «bad» in American English; the vowels should be notably different. The former is nasalized and slightly raised.

André G. Isaak

Also, compare those with «can’t», which for many Americans is pronounced [kʰæ̃t] with no [n] at all. If your native language is one which lacks phonemic nasal vowels, vowel nasalization is often heard as an [n] even when none is actually present.

John Savard

André G. Isaak said:

If your native language is one which lacks phonemic nasal vowels,

And, of course, English is such a language. (Or, at least, the dialects thereof with which I am most familiar.)

Peter Baker

I'm trying to take account of all the vowel + rhotic hook combinations I've seen. Mostly this is not difficult: ɝ, ɚ and a few others can be treated as ligatures, while for some kerning gives a good result.

The most difficult combination is ɤ (ram's horns) + rhotic hook. When I've seen the combination online it's just been the two characters with no attempt to connect them. Following the recommendations here, with no kerning, I get this:

Creating a ligature is not much better:

I don't have any more thoughts about this. Do people here have any suggestions?

John Hudson

Something like this?

Igor Freiberger

My full set of vowels with rhotic hook:

Peter Baker

John: I like that a lot. Thanks!

Igor: these are very elegant. Are they all used by linguists, or are you after supplying a complete set just in case?

Igor Freiberger

Thanks, Peter. The set is based on what I found in a number of linguistic-aware fonts. I am trying not to use the "just in case" approach because the font is already very large (10k+ glyphs). But I am not a linguist so the set may be larger than the real need.
 
This font family should be released next February (finally!). It includes a large phonetic set aimed to cover almost all linguist needs.



The image does not show the phonetic glyphs that are part of "regular" Latin script, like ɑ, nor combining or modifier glyphs.

Peter Baker

Nice, Igor! And better a few too many, I guess, than not meeting the need.