Font needs reworking

[simoncozens]

This font has some significant issues:

• It doesn't reorder properly (#4). This is because it tries to handle reordering manually, when these days a better solution is to use the Universal Shaping Engine to do the reordering for you.
• Because it handles manual reordering, it contains a large number of ligature glyphs which (when converted to USE) are completely superfluous.
• There are limited anchor attachments (#2).
• There are a number of missing ligatures and substitutions (#5).

These are all shaping issues, and I could conceivably fix them. However, there are also some fundamental drawing issues as well:

• Tai Tham is based on a circular model (like Myanmar), whereas this font is condensed in its default position.
• Some glyphs (for example, ᨫ, and the subjoined form of nga e.g. ᨡ᩠ᨦ) have the wrong shape.

Given that we are hoping to have a reworking on Noto Sans/Serif Myanmar, it makes sense for us to consider Tai Tham part of the same project. As such, I don't plan to work more on these issues in isolation, but we will get them fixed when we rework the font.

[ohbendy]

Just to note that Tham as used by Khuen is a circular script like Myanmar; Tham Lanna may differ a bit with more dynamic proportions, loops and folded terminals:

The Noto Tham font will need to decide whether it's for Lanna, Isaan-Lao, Khuen or Lue since they each have their own letterforms and layout requirements.

[simoncozens]

I think we'll want Khuen style initially.

[r12a]

Some time ago (+3 for the pandemic) i attended some Unicode meetings that discussed changing the encoding model for Lanna. I'm not sure what the current status is, as i haven't heard anything for a while, but i believe it would have involved changes to the USE. Maybe someone reading this has more up to date information, but if not i could ping Debbie Anderson to ask about the current status.

Note, btw, that this is one of those scripts where a few characters used in Khün are not used in Northern Thai, and vice versa, and to apply the same letterform styling across the board will probably look wierd if you have Khün styling and use it for Northern Thai. Northern Thai also tends to stack 3 items, whereas Khün tends to stick at 2. I suspect that ideally we'd have 2 fonts.

Northern Thai

Khün

[r12a]

Here's a list that shows significant deviations in letterform between Northern Thai and Tai Khün in a typical font.

The most significant difference is the last item, ᩋ U+1A4B TAI THAM LETTER A. This is frequently used - eg. it is used for standalone vowels, and the shape is altogether different, not just a styling quirk. Afaik it's not appropriate to just use either shape in a both orthographies.

This is a bit like Syriac, which was recently split back into 3 fonts.

[r12a]

In case it helps, here's a little more information about the shapes used for the Noto Tai Tham font.

Rounded vs not-so-rounded

The Noto font uses rounded shapes, usually associated with Khün where the Northern Thai shapes tend to be slightly more complicated. This includes the following letters.

ᨡ U+1A21 TAI THAM LETTER HIGH KHA ᨩ U+1A29 TAI THAM LETTER LOW CA ᨬ U+1A2C TAI THAM LETTER NYA ᨯ U+1A2F TAI THAM LETTER DA ᨪ U+1A2A TAI THAM LETTER LOW SA

Noto font:

A Tai Tham KH New font (Khün style):

Payap Lanna font (NThai style):

Hariphunchai font (NThai style):

Substantially different

Two of the 3 shapes that tend to differ significantly are the following.

ᨫ U+1A2B TAI THAM LETTER LOW CHA ᨨ U+1A28 TAI THAM LETTER HIGH CHA

Noto font:

A Tai Tham KH New font (Khün style):

Payap Lanna font (NThai style):

Hariphunchai font (NThai style):

LETTER A

ᩋ U+1A4B TAI THAM LETTER A

This is the third letter that has a substantially different shape in the two writing styles. What's worth noting here is that the Noto font uses the form usually associated with Northern Thai, whereas the shape of the other characters were more similar to what is often used for Tai Khün. This seems an oddity.

Noto font:

A Tai Tham KH New font (Khün style):

Payap Lanna font (NThai style):

Hariphunchai font (NThai style):

Of course, besides differences in basic letter shapes, of which these are examples, there are also differences in shape and occasionally behaviour for subjoined consonants and combining marks.

By the way, i think an advantage to having 2 separate fonts for Tai Tham (like we have for Syriac) will also help browsers select appropriate generic styles for font fallback. This is being worked on in CSS at the moment.