You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
Jihvamuliya and upadhmaniya do not stack in Devanagari #4
Noto Sans Devanagari: Version 2.000;GOOG;noto-source:20170915:90ef993387c0
Noto Serif Devanagari: Version 1.01 uh
Issue
As of Unicode 11.0, U+1CF5 VEDIC SIGN JIHVAMULIYA and U+1CF6 VEDIC SIGN UPADHMANIYA have Indic_Syllabic_Category=Consonant_With_Stacker, so the following consonant should be subjoined. Cf. notofonts/brahmi#7.
Character data
ᳵकᳵखᳶपᳶफ
U+1CF5 VEDIC SIGN JIHVAMULIYA
U+0915 DEVANAGARI LETTER KA
U+1CF5 VEDIC SIGN JIHVAMULIYA
U+0916 DEVANAGARI LETTER KHA
U+1CF6 VEDIC SIGN UPADHMANIYA
U+092A DEVANAGARI LETTER PA
U+1CF6 VEDIC SIGN UPADHMANIYA
U+092B DEVANAGARI LETTER PHA
The background documents are L2/17-319 and L2/17-095.
These glyphs are mainly and widely used in Bengali script. These forms are also occasionally seen in few Devanagari books published in Bengal ( See the last page of L2/17-319 where the non-stacked forms are used)
In manuscripts and inscriptions of native Devanagari regions like Gujarat/Rajasthan, Maharashtra, Madhya Pradesh, Uttar Pradesh the forms in L2/17-095 are used which are not yet encoded and which also can have stacked conjuncts.
However the stacked forms need to supported in Bengali fonts, may not be needed in Devanagari.
Because this font has been designed with precomposed glyphs, rather than composable marks, it looks like supporting the stacked conjuncts will require a lot of extra drawing. One question I have is whether the four cases mentioned in L2/17-095:
Jihvamuliya ೱ + Ka क
Jihvamuliya ೱ + Kha ख
Upadhmaniya ᳶ +Pa प
Upadhmaniya ᳶ +Pha फ
are exclusive, or if stacking would need to apply to other bases.
"Stacking" does not always occur with jihvamuliya and upadhmaniya special signs (note that Unicode charts display them within a dashed box (meaning that they are more like special format modifiers) and not with the dotted circle used by other diacritics.
In fact these signs may also occur:
before on the left of a base consonant cluster (their encoding in that case should be followed by a ZWNJ), or
in a ligature attached to the left of a base consonant cluster (their encoding in that case should be followed by a ZWJ).
The absence of ZWNJ or ZWJ joiners will move them above (not below!) the first consonant in the cluster, which may still be followed by other combining diacritics (notably other consonant signs, invisible viramas between consonants rendered in half forms, visible virama signs after the last consonant of the cluster, dependant vowel signs that may stack averywhere around the cluster and length marks), but in some complex clusters.
This concerns Devanagari, but also Sharada, Marathi, Bengali-Assamese, Kannada, and a few other Brahmic scripts, that use such adapted signs for transcription of Sanskrit sounds inherited from the Brahmi script (where the signs were simply plain letters similar to consonants).
The forms in Noto sans Devanagari are proper without stacks and can be retained as such without modifications ,
Those glyphs from Vedic extns are primarily seen in Bengali.
For the forms from Devanagari mss/ inscriptions it depends on UTc decision if they have to be represented using the code points in Vedic Extns block.
For now the both Serif and Sans fonts can be made identical
Activity
sridatta1 commentedon Jun 7, 2018
The background documents are L2/17-319 and L2/17-095.

These glyphs are mainly and widely used in Bengali script. These forms are also occasionally seen in few Devanagari books published in Bengal ( See the last page of L2/17-319 where the non-stacked forms are used)
In manuscripts and inscriptions of native Devanagari regions like Gujarat/Rajasthan, Maharashtra, Madhya Pradesh, Uttar Pradesh the forms in L2/17-095 are used which are not yet encoded and which also can have stacked conjuncts.
However the stacked forms need to supported in Bengali fonts, may not be needed in Devanagari.
The glyph needs to changed for Upadhmaniya https://www.unicode.org/versions/Unicode11.0.0/erratafixed.html
marekjez86 commentedon Oct 2, 2019
fixed in https://github.com/googlefonts/noto-fonts/tree/master/phaseIII_only/unhinted/ttf/NotoSerifDevanagari
marekjez86 commentedon Oct 2, 2019
reopening for Sans (my bad)
nizarsq commentedon Jun 8, 2020
Current rendering in NotoSansDevanagari-Regular.ttf and NotoSerifDevanagari-Regular.ttf

simoncozens commentedon Feb 15, 2023
Because this font has been designed with precomposed glyphs, rather than composable marks, it looks like supporting the stacked conjuncts will require a lot of extra drawing. One question I have is whether the four cases mentioned in L2/17-095:
are exclusive, or if stacking would need to apply to other bases.
virtualvinodh commentedon Apr 2, 2023
@simoncozens AFAIK Grammatically speaking, they cannot appear before other consonants.
V
verdy-p commentedon Jun 18, 2025
"Stacking" does not always occur with jihvamuliya and upadhmaniya special signs (note that Unicode charts display them within a dashed box (meaning that they are more like special format modifiers) and not with the dotted circle used by other diacritics.
In fact these signs may also occur:
This concerns Devanagari, but also Sharada, Marathi, Bengali-Assamese, Kannada, and a few other Brahmic scripts, that use such adapted signs for transcription of Sanskrit sounds inherited from the Brahmi script (where the signs were simply plain letters similar to consonants).
sridatta1 commentedon Jun 20, 2025
The forms in Noto sans Devanagari are proper without stacks and can be retained as such without modifications ,
Those glyphs from Vedic extns are primarily seen in Bengali.
For the forms from Devanagari mss/ inscriptions it depends on UTc decision if they have to be represented using the code points in Vedic Extns block.
For now the both Serif and Sans fonts can be made identical
sridatta1 commentedon Jun 20, 2025