free font with no dot on i

[Quoth]

Why do I want a font with no dot on the i for ebooks?
Because traditionally Irish used no dot on the i, because some are í. Additionally some letters had a dot to indicated hard/soft/accent which are now shown by suffix of an h: bh, dh, gh, mh, etc.
Georgia and some other fonts do have a dotless i: ı, but that's only good for paper as not in the i position. I want one that has ı for the i position to allow fall back to i rather than [] in the text!
Paper is no problem, because you include the fonts in the PDF and that's exactly what is used.
I suspect old Scottish is the same as it has ì and í in the Gaelic. Irish only has the áéíóú as vowel accents, so those are not a problem.
Some old Irish words lost the dots before dots were changed to an added h, so the name Badb is really Badhbh, though that might not help people much. It might be close to Baeyf or Baeyv for an English speaker.

It's also good if the a, g, n and & etc are in the older Gaelic styles. Both modern Gaelic (and old stuff) and Hebrew use a symbol like a cross between a 1 and 7 for the & symbol which in English is Et and used to be listed after z. The Irish is closer to a 7 than a 1 and the Hebrew & is actually their letter Vav.

[Jellby]

Do you want just the dotless i or do you want a full Gaelic feel? For the latter check e.g. the external links in https://en.wikipedia.org/wiki/Gaelic_type

I don't know if any non-decorative font has a dotless i by default, or maybe as an OpenType option, but it should be relatively easy to modify a font and remove the dot. There's also a specific dotless i character that is used in Turkish at least, you could use any font that supports it and replace all your i's with ı's (although it will probably break searching, etc.) If modifying a font, you could just replace the "i" glyph with the "ı", which should be even simpler than removing the dot.

You may find what you want searching for fonts for scholars or medievalists, like Junicode. Apparently Alphabetum offers a stylistic alternate for dotless i.

[Tex2002ans]

You may also want to check out the fonts mentioned in this Source Code Pro bug:

#205: "ḃ and ḋ (b & d with dot above)"

Some of the users mentioned 10 fonts with those Gaelic characters. Perhaps they also have a stylistic dotless i.

Originally Posted by Jellby

There's also a specific dotless i character that is used in Turkish at least, you could use any font that supports it and replace all your i's with ı's (although it will probably break searching, etc.) If modifying a font, you could just replace the "i" glyph with the "ı", which should be even simpler than removing the dot.

Searching for some more info on this, I stumbled upon this answer in the Unicode Mail List:

"Re: Irish dotless I"

Irish has no distinction between a dotted or undotted glyph representation of the vowel i...

So which ever character is encoded does not matter as they will be perceived as equivalent by Irish readers. If one creates a Irish document using the dotless i, it won't break the Irish orthograph, and there's a way to use a custom folding rule for Irish to use any of these two characters.

Because of that, an application may as well use a Gaelic font where the soft-dotted 'i' is shown without the dot, or mutate every soft-dotted 'i' into dotless 'i' before rendering with any other font.

In both cases, there's apparently no need to add a new character or diacritic to preserve a linguistic distinction, or to render Irish with either the modern or traditional Gaelic styles...

And the Wikipedia page on "Dotted and dotless I":

Usage in other languages

[...]

The dotless ı may also be used as a stylistic variant of the dotted i, without there being any meaningful difference between them. This is common in Irish, for example, but is considered simply an omission of the tittle rather than a separate letter.

Based on this, I wouldn't insert the actual dotless character.

Perhaps as Jellby said, apply the new style via OpenType, or create an embedded font with the dotless overwriting the normal 'i' slot.

Side Note: This solution is probably the right one for Text-to-Speech (TTS) as well.

Originally Posted by Quoth

Both modern Gaelic (and old stuff) and Hebrew use a symbol like a cross between a 1 and 7 for the & symbol which in English is Et and used to be listed after z. The Irish is closer to a 7 than a 1 and the Hebrew & is actually their letter Vav.

You've read the fantastic "Shady Characters" book/blog, correct?

"The Ampersand, Part 1 of 2"
"The Ampersand, Part 2 of 2"
"The Ampersand, Part 2½ of 2"

[Quoth]

No, I'd not insert any dotless i character such as ı that's not at the i position in the font.
I'd read the Wikipedia article and various ones on Et / & / 7, thanks.
I've got a load of Gaelic fonts and some that are Celtic styles. Most of the more accurate ones don't seem to work in even the most lax epub viewer and some not even in LibreOffice writer. We used a Bunchló font years ago and it's less than ideal. I don't use it now.

I think the best solution might be to edit and rename some free fonts so the ı is at the i position. Then search, text to speech and font fallback works. Similarly with the old insular styled r, s, d, f, g, t, though often the a, d, f g and t are replaced by the insular/ old Irish styles in the regular position. Using a font where the old style r, s, d, f, g, t, and dot-less i are additional characters and not replacing the regular ones causes [] on fall back, difficulty in search (can't be entered!) and issues with some text to speech.

However ebooks in the past have been worse for non-latin-roman other than a few extra characters for many versions of kindle (no Asian, full Greek, Arabic, Hebrew or Cyrillic) and it was a solved problem maybe 15 years earlier. Typical careless US-centric software.

It's purely for stylistic effect in a few passages. I'm not going to publish actual complete old Irish or Scottish texts! My family members that are fluent in Irish can't easily read pre 1948 Irish! The 16th C or older (old Irish, never mind 11th C Insular) is probably harder for a native Irish speaker than Chaucer for English speakers, though I find it helps a little to read Chaucer out loud.

[Quoth]

Interestingly some place names near here in Irish have the letter n in them and it's thought originally it was an r. You'd have to see old Irish script to see how that's possible.

Mind your Ps and Qs
Thought to apply to crossing the Irish sea. Welsh replaced the older Q sound at the end of a word with the P sound. In modern Irish a c. Like ancient Latin and Anglo-saxon, C is usually hard. Irish mac = Welsh map = son in English. The Welsh dropped the m. Anglo-Saxon Cween was changed to Queen by the Normans.

[Jellby]

I knew I had done this, but didn't remember the dotless i specifically. Take a look at this book (HiTech, vol 1, chapter 20, maybe others). The trick there is the "ss02" stylistic set of Junicode (see here), but I guess that's too much for you, it looks nice though.

[Quoth]

I might read that anyway! Thanks.
Buaḃall bionngloraċ go mbuaiḋ ninnscne ⁊ nurlaḃra, ceille, ⁊ coṁairle, go ttaidḃrēḋ seirce ina ḋreiċ attar lá gaċ aen at as cíoḋ

now to be archived https://sourceforge.net/projects/junicode/files/
because there is now a newer version of junicode under development, but not yet finished.
https://github.com/psb1558/Junicode-New
I'll look at old and new and then test on ereaders that support publisher fonts and don't support them.

Looks good for Anglo-Saxon too.

[Jellby]

Actually, I realized that the ⁊ was not needed, since ss02 already transforms & into ⁊

[Jellby]

Originally Posted by Quoth

there is now a newer version of junicode under development, but not yet finished.
https://github.com/psb1558/Junicode-New

Ah, and there is a test page (https://psb1558.github.io/Junicode-New/) where you can enable just the dotless i (see in "Show more", cv13[1]), with the downside that it also turns I into a dotted I.

[Quoth]

My Irish speakers had to transcribe that to modern Irish to read it.

Buaḃall bionngloraċ go mbuaiḋ ninnscne ⁊ nurlaḃra, ceille, ⁊ coṁairle, go ttaidḃrēḋ seirce ina ḋreiċ attar lá gaċ aen at as cíoḋ

BECOMES

Buallaidb bionn glórach go mbuaigh n-innscne agus n-urlabhra, ceille, agus comhairle, go t-aidbhreidh serce ina dhrecich atá ár lá gach aon áit as cíodhh

Even so it's got archaic words and a wrong gender. Also aen is probably aon.

It bears loudly that gender and speech, sense and counsel, win the admiration of affection in which our day is everywhere.

or if you transliterate a little differently:

It's is loud to hear that ingenuity and speech, understanding and advice, that admiration of affection is a genius in which our day is everywhere.

7 = agus = and = &

Irish has hugely changed in the last 150 years, whereas 200 year old English is fine. Manx is worse as they anglicised the spelling!

[Quoth]

Originally Posted by Jellby

Ah, and there is a test page (https://psb1558.github.io/Junicode-New/) where you can enable just the dotless i (see in "Show more", cv13[1]), with the downside that it also turns I into a dotted I.

That's odd.

[Tex2002ans]

Originally Posted by Quoth

now to be archived https://sourceforge.net/projects/junicode/files/

because there is now a newer version of junicode under development, but not yet finished.

https://github.com/psb1558/Junicode-New

Thanks for this info. I wasn't aware Junicode was archived.

I recently went through a few fonts I've downloaded over the years and was pleasantly surprised to see many updates.

Originally Posted by Quoth

I'll look at old and new and then test on ereaders that support publisher fonts and don't support them.

You may also be interested in these two topics:

2020: "Forcing old-style figures from a font in Kindle"

2019: "Turn off ligatures (temporarily)?"

There was a lot of discussion around OpenType features using different methods and devices.

[Quoth]

Originally Posted by Tex2002ans

Thanks for this info. I wasn't aware Junicode was archived.

Not yet, it will be an old version when the replacement is finished.

Final version from

Buaḃall bionngloraċ go mbuaiḋ ninnscne ⁊ nurlaḃra, ceille, ⁊ coṁairle, go ttaidḃrēḋ seirce ina ḋreiċ attar lá gaċ aen at as cíoḋ

It is loudly heard that ingenuity and speech, wisdom and counsel admirably wins affection in the genius which is in our day everywhere without exception.

But they are not confident.

Also there is curiosity as to WHY Twain and Warner had what must be an Irish quotation and where did they get it?

[Quoth]

Junicode does "work" on my system, but I'd have to edit it and swap the insular letters and the regular basic Latin so that the fallback is the ordinary letters and that search and text to speech both work, as you can only enter a limited range of letters on the built in keyboards.

So without editing it, it's useless for universal ebooks. Fine for paper or epub ereaders with good embedded font support but paper via PDF isn't a problem with the fonts I already had that are easier to use.

[Jellby]

Originally Posted by Quoth

Also there is curiosity as to WHY Twain and Warner had what must be an Irish quotation and where did they get it?

Did you check the other chapters? It has a few quotations in much more obscure languages/scripts. I think it was a kind of parody on "serious" books that find it necessary to insert quotations in (typically) Latin or Greek. I haven't read the book, though.

Originally Posted by Quoth

Junicode does "work" on my system, but I'd have to edit it and swap the insular letters and the regular basic Latin. [...] So without editing it, it's useless for universal ebooks.

Huh? The fallback should be basic Latin. If you write "this & that", that's what you'd see. It's only if you activate a feature such as ss02 that is displayed with some more fanciness, and you can do that for specific words/paragraphs/sections with CSS.

If you want only some of the substitutions activated by ss02 and not others, you'd have to edit the font indeed. But instead of switching characters, you could just edit the ss02 table (or create a new one).

If it's just a stylistic choice and no big deal if a particular reader does not support OpenType features, then you're done. Good readers will display what you want, poor readers will display the readable base Latin characters, and search and TTS should work in both. If you require that the alternate characters must be displayed, then you'd need to manually switch glyphs, but you're doomed anyway, because not all readers (both human and device) like/support/enable embedded fonts.