NRSI: Computers & Writing Systems
SIL Unicode Roman Fonts – FAQ and Known Issues
All the information on our Roman fonts have been moved to a new web site. Please click here for the new site.
SIL Unicode Roman Font Pages
The content below is old. Please click here for latest fonts and FAQs on the new site.
‘I have already downloaded and begun to use these fonts (Charis SIL and Doulos SIL). They are a wonderful achievement, not only because of their comprehensiveness of design but also because of their flexibility of use — this latter aspect immeasurably enhanced by the Open Font License attached to them. To release these products of such excellent quality under SIL’s OFL is an act of generosity for which the world of international verbal communication has reason to be eternally grateful.’
Table of Contents for this web page:
Frequently Asked Questions
About the font / Using the font
Question: Why is there an inversion of names? (eg SIL Doulos -> Doulos SIL and SIL Charis -> Charis SIL)
Answer: If SIL is in front of a font name then that probably means it is a legacy font. If it is after the font name it probably means it is Unicode.
Question: Are the SIL Unicode Roman fonts going to stay free?
Answer: There is no intention to ever charge users for using the SIL Unicode Roman fonts. The current versions are licensed under a free/open license and future versions will be similar.
Question: Do I still need to use the “SIL Unicode IPA font beta” (SILDoulosUnicodeIPA) font or can I just use “Doulos SIL” or “Charis SIL”?
Answer: Everything in the “SIL Unicode IPA font beta” is included in the SIL Unicode Roman fonts so you no longer need the IPA font.
Question: How do I type IPA characters using your Roman fonts?
Answer: To type IPA characters you will need to download and install an IPA keyboard from the IPA keyboards page. Keyboards are available for Windows, Mac, and Ubuntu (Linux) machines.
Question: I need to distinguish between single-storey 'ɑ' and double storey 'a' in italic IPA text, but the font doesn't allow it.
Answer: This can be addressed using font features.
This overview explains what font features are and what each feature does. (The feature that will allow you to display an italic 'a' is number 1053 - Slant Italic Specials.)
This page explains how to activate font features in various applications/operating systems.
Question: I would like to bundle one of the SIL Unicode Roman fonts with my application - can I?
Answer: The SIL Open Font License allows bundling with applications, even commercial ones, with some restrictions.
See the OFL web page.
Question: Can I use one of the SIL Unicode Roman fonts on my web site?
Answer: You can certainly create web pages that request one of the SIL Unicode Roman fonts be used to display them (if that font is available on the user's system). According to the license, you are even allowed to place the font on your site for people to download it. We would strongly recommend, however, that you direct users to our site to download the font. This ensures that they are always using the most recent version with bug fixes, etc.
Question: Will documents created with earlier (legacy) fonts such as the SIL IPA and IPA93 fonts be compatible with the new (Unicode) version?
Answer: No, documents which were created (encoded) with legacy fonts are not compatible with Unicode fonts. You will need to convert your data to Unicode. You can use TECkit for this process. We have mapping files (which work with TECkit) for converting documents which used SIL’s IPA fonts to Unicode. For instructions, see SIL IPA93 Data Conversion.
Linespacing and font metrics
Question: Why is the line spacing so much looser than other fonts, such as Times New Roman or Gentium?
Answer: Our SIL Unicode Roman fonts include characters with multiple stacked diacritics that need a much looser line spacing (for example, U+1EA8 ). We cannot make the line spacing tighter without experiencing “clipping” of those characters. You may be able to overcome this by adjusting the line spacing in the application. For example, in Microsoft Word select and set the line spacing to use the setting and a value more suited to your needs. For example, if the font size is 12 pt, select line spacing of 13 pt. This will give a tighter line spacing. You can adjust the value up or down depending on how many diacritics you need to stack. With HTML you should also be able to change the line spacing; add the line-height property to your tag (“line-height:105%;” or “line-height: 12pt;”) and play around with the value until you get the spacing desired.
With version 4.106 of our fonts, we have provided for download "compact" versions of our fonts (ie Charis SIL Compact and Doulos SIL Compact). These fonts provide linespacing similar to the linespacing in Times New Roman.
Question: Will font and glyph metrics stay the same in future versions?
Answer: We do not guarantee to keep metrics stable in future versions. The practical result of this means that you should expect to have different line lengths, paragraph length may be different, and line spacing may even be different. You should not expect your document to have the same page layout as you do with the current font.
Character inventory / Glyph shapes
Question: Why do the SIL Unicode Roman fonts have some Greek characters, but not all?
Answer: While it is true that the font includes some Greek characters, it is not intended to provide general support for the Greek language. Those Greek characters that were included were done so in order to support various (primarily linguistic) notational systems. If Greek language support is needed, the Galatia SIL and Gentium fonts are two available options.
Question: It seems like U+027F LATIN SMALL LETTER REVERSED R WITH FISHHOOK and U+0285 LATIN SMALL LETTER SQUAT REVERSED ESH are not correct. Why is the shape different than what is in the Unicode book?
Answer: In the Unicode book U+027F LATIN SMALL LETTER REVERSED R WITH FISHHOOK and U+027E LATIN SMALL LETTER R WITH FISHHOOK are the same height and shape, just mirrored. Our understanding is that the U+027F LATIN SMALL LETTER REVERSED R WITH FISHHOOK and U+0285 LATIN SMALL LETTER SQUAT REVERSED ESH glyphs in our SIL Unicode Roman fonts are correct and the ones in the Unicode book are wrong. As used in China the stem of U+027F LATIN SMALL LETTER REVERSED R WITH FISHHOOK and U+0285 LATIN SMALL LETTER SQUAT REVERSED ESH always extend below the baseline. We have brought this to the attention of the Unicode Technical Committee (see: Glyph corrections for U+027F and U+0285 in TUS) and Microsoft and we believe changes will be made in the future.
Modifying the font
Question: I would like to modify one of the SIL Unicode Roman fonts to add a couple of characters I need. Can I?
Answer: Yes - that is allowed as long as you abide by the conditions of the SIL Open Font License.
Question: So will you add glyphs to one of the SIL Unicode Roman fonts upon request?
Answer: If you have a special symbol that you need (say, for a particular transcription system), the best means of doing so will be to ensure that the symbol makes it into the Unicode Standard. It is impossible for us to add every glyph that every person desires, but we do place a high priority on adding pretty much anything that falls in certain Unicode ranges (extended Latin, Cyrillic). You can send us your requests, but please understand that we are unlikely to add symbols where the user base is very small, unless they have been accepted into Unicode.
Question: Can I send you work I've done to be incorporated into the SIL Unicode Roman fonts?
Answer: Yes! See the FONTLOG for information on becoming a contributor.
Question: I'm having problems making PDFs — why won't my document distill?
Answer: The SIL Unicode Roman fonts are large fonts, with lots of glyphs. As a result, some printers can balk at PDFs that have the complete font embedded. The easiest way to avoid this is to have Acrobat/Distiller subset the font. This is generally a good idea anyway (with any font) and can reduce the size of your files.
Mac OS X
Question: Can the SIL Unicode Roman fonts be used with Word 2004 on Mac OS X?
Answer: Since Word 2004 is Unicode-based, and the SIL Unicode Roman fonts are Unicode fonts, you would expect to be able to use them with Word 2004. And you can — to a point. The SIL Unicode Roman fonts rely on Uniscribe, Graphite and AAT “smart rendering” technologies to position diacritics, contour tone letters, handle ligatures, etc.
Microsoft has not implemented "smart rendering" in Word 2004, and therefore our SIL Unicode Roman fonts will not position diacritics properly, contour tone letters, or handle ligatures.
So, the combination of Word 2004 and our SIL Unicode Roman fonts is not a complete solution. Whether it’s adequate for you depends on whether you need the capabilities that are missing due to the lack of smart rendering in Word 2004. You may wish to use one of the applications listed here (Applications that provide an adequate level of support for SIL Unicode Roman fonts) instead.
Rendering: Combining marks/Ligation/PUA
Question: How do I tell what version of Uniscribe I’m using? I’ve been told I have to have Uniscribe 1.0468.4015.0 (main.030328-1500) or later for accurate diacritic positioning. I put usp10.dll in with notepad.exe in a directory, but I am not sure that it is actually using that Uniscribe.
Answer: The key is a program called msinfo32.exe. It certainly will be on your machine if you have MS Office, but may be provided in other configurations. On some machines it is in C:Program FilesCommon FilesMicrosoft SharedMSInfomsinfo32.exe. (This is the app that is launched if you click the button on an Office application’s button.)
First, launch msinfo32.exe. Under , select . It will take a bit to load the list. Then scroll down looking for usp10.dll. You may see it loaded several times, from several different directories. But if you haven’t yet launched your special copy of Notepad.exe, then you probably won’t see that directory mentioned. Now launch your Notepad and then refresh the display — you should see a usp10.dll loaded from your directory — a sure sign that Notepad is using the local copy.
Question: I am using Word 2003 and some of the diacritics are not shown, although they are there (as can be proven with the Show Unicode Macro of the UnicodeWordMacros.dot and also when I copy/paste data from Word to Notepad). What is going on?
Answer: Check to see if is set. If you do not have a tab under , you should:
Question: I noticed that when I put a cedilla under some characters it renders it as a “comma”. When I do “Show Unicode”, it gives me the same Unicode codepoint for both, so it is just a rendering issue. Is this intentional?
Answer: A careful study of the Unicode repertoire shows that, for example, character U+0157 LATIN SMALL LETTER R WITH CEDILLA (which decomposes to <0072 + 0327> typically is drawn with the comma-shape rather than cedilla shape. This happens for a number of characters, including g/G, k/K, l/L, n/N, and r/R. (Interesting aside: notice that for lower case g the cedilla, drawn as a comma mark, is actually rendered above the g)
Additionally, s/S and t/T with cedilla are sometimes rendered with the comma shape — thus we have a feature in the Graphite code (Romanian Style) and language-specific behavior in the OpenType code (attached to language “Romanian”) that cause these combinations to be rendered with the comma style. (This alternate rendering predates the introduction into Unicode 3.0 of s/S and t/T with comma below [U+0218, U+0219, U+021A and U+021B] which are now the preferred way to distinguish these characters).
Question: When I select the Vietnamese language in Microsoft Word I am not seeing Vietnamese-style diacritics. Is this a bug in the font?
Answer: In Microsoft Word it appears that you must turn on "Ligatures" in order to get the Vietnamese diacritic positions. Select your text (or your style) and then go to and select . This should fix the problem.
Question: Why don’t my tonebars ligate?
Answer: See “Why don’t my diacritics position properly?”
Question: Why is there space between my tonebars in Word?
Answer: In Microsoft Word you must turn on kerning in order for the Tonebars to position correctly. Go to and set to be
Question: When I type data, I get the proper characters, but the stacked diacritics show up on top of one another, rather than stacked, and not properly centered. What am I doing wrong?
Answer: See “Why don’t my diacritics position properly?”
Question: Why don’t my diacritics position properly?
Answer: Cause 1: The application you are using is neither Graphite-aware nor OpenType-aware, or your Uniscribe needs to be updated, or you are using characters from the Private Use Area (PUA) in a Uniscribe-based application.
In order for complex behaviors such as diacritic positioning and ligatures to work, the application must be able to use the Graphite or OpenType tables in the font. For example, until Microsoft Office 2003 was released there were no versions of Microsoft Office and of the system component Uniscribe that had the ability to use either of these for Latin script. Unfortunately, even the latest versions of Uniscribe ignore OpenType information for characters from the PUA area, so neither diacritic positioning nor ligation occur. Microsoft says this is by design.
Cause 2: If you are using Word 2000 or Word XP with an updated version of Uniscribe, some kinds of display problems can be fixed by saving and reopening the file.
Cause 3: While some of these problems are font errors (that we want to know about), another common cause is formatting issues within the application. In order for diacritic positioning or ligatures to work correctly, the application must render the complete character sequence in one operation. The most common reason for this condition to fail is if some characters in the sequence have different formatting than the others. If there is any difference at all in the formatting (e.g., in character spacing or color, font names or sizes, etc.) the application may have to break the sequence into separate runs.
Solution: In order to rule out formatting problems, make absolutely sure that the characters in the sequence are formatted identically. Some applications let you copy the affected text to the clipboard and then use to paste unformatted text back into the document. Another approach, available in Microsoft Word, is to select the text and press Ctrl – space to reset all character formatting to the paragraph default. (This assumes your default paragraph style is formatted with one of our SIL Unicode Roman fonts).
In Word, even formatting such as Complex Scripts font and Asian Text font settings must match exactly for the entire sequence, even though these settings aren’t actually used to render Latin text.
Cause 4: Adobe applications (and thus InDesign) do not yet handle dynamic diacritic placement.
Question: Why don’t my diacritics positioned above or below characters appear onscreen, but they do show up in print?
Answer: Solution: The vertical metrics for this font have been set to accomodate the majority of situations, but in some scenarios, especially with stacking diacritics, you may get clipping onscreen. You may be able to overcome this by adjusting the line spacing in the application. For example, in Microsoft Word select and set the line spacing to use the setting and a value approximately twice the font size. For example, if the font size is 12 pt, select line spacing of 24 pt. You can adjust the value up or down depending on how many diacritics you need to stack.
Question: Why don’t the PUA characters work properly (diacritic positioning, tone ligation, etc)?
Answer: Uniscribe ignores complex behaviors that have been provided in fonts for PUA characters, and thus in Uniscribe-based applications such as Paratext 6 and Microsoft Word the PUA characters will not display correctly.
Question: In certain combinations, two upper diacritics (e.g., tilde over macron) display in a fixed order (the tilde below the macron), no matter what order they are typed in. Why is that?
Question: Why don’t the overlaid combining marks render properly in the font?
Answer: See “The font does not support some combining marks”.
Question: Why don’t some of my characters render in Internet Explorer?
Question: Why are some of my diacritics colliding with nearby letters?
Answer: When combined with some narrow glyphs (such as ’i’), wide diacritics (such as the tilde) may collide with adjacent glyphs. In many cases this is not a problem (it is sometimes OK for glyphs to collide). If this causes difficulty with the legibility of the text, then manually space those letters apart in your text using manual kerning or character spacing settings in your application. We do not have a generally feasible solution for this problem, but will continue to look for one.
Question: How do I use a feature? For example, I see there are four Eng ( ) variants. How do I choose which variant displays?
Answer: The answer depends on the application in question:
For more information and examples, see XeTeX documentation and sample files.
So, anticipating your (or someone’s) next question: What do I do if I’m using Word or other Uniscribe-based apps?
Question: How do I use the Small Caps feature?
Answer: The Small Caps feature is an OpenType and a Graphite feature that can be turned on within the font. How to use it will vary from one application to the next.
Onscreen display messed up on Windows XP
Some of the letters appear too close together or too far apart. With version 5.x of our Roman fonts we implemented a new hinting strategy. The new hinting strategy effectively has no x-direction hints, since all current OSes and browsers ignore those hints anyway. The trouble here is specifically with Windows XP.
If it looks too bad, you may choose to stick with version 4.x of our fonts. The problem should disappear when you print to a printer or to pdf.
Performance hit in OpenOffice 3.2
There is a Known Issue with regard to using Graphite-enabled fonts in OpenOffice 3.2. See Graphite in OpenOffice.org for a solution.
The Graphite features in these fonts are not currently handled by the CSS support in Firefox 11.
Application crashes when printing
People have been experiencing problems when printing documents using our fonts (also failed using Word 2007 SP2's ability to save as pdf). This is apparently a print spooler problem. Microsoft has resolved this issue as of 13 January 2010 through Windows Update. If you do a Windows Update you should make sure you install KB972270. Read about it here http://www.microsoft.com/technet/security/Bulletin/MS10-001.mspx. If you "hid" the previous update (KB961371), you do not need to "unhide" it. KB972270 completely replaces KB961371.
Old answer (2009-08-19): There are two solutions. One is to disable the print spooler and print directly to your printer. This may not be possible on a network printer. The other solution is to uninstall Microsoft Update KB961371 (see http://support.microsoft.com/kb/961371). Do this at your own security risk! You will need to go to . Make sure is selected. Scroll almost to the bottom where you see "Windows XP - Software Updates" and choose "Security Update for Windows XP (KB96137)" (it should be dated around the middle of July 2009, but this will depend on when you installed the update). Click on . You may have to reboot. In order to prevent installation of this update you will need to go through Windows Update again, select and "hide" this update. We will continue to research this issue to find a better solution.
Line metrics from legacy fonts cannot be imported using TypeTuner
This was a problem with version 4.106. We updated the fonts (without changing the version) and the current download (as of May 5, 2009) corrects the problem. Please download the fonts again if you had this problem.
Small caps feature
In the Graphite version of OpenOffice the Vietnamese feature PLUS Small Caps and the Ogonek feature PLUS Small Caps do not work properly (the base character is turned into a small cap, but the diacritics do not follow the other feature). The combination of features work properly in FieldWorks applications, including WorldPad. This is because FieldWorks uses NFD and OpenOffice uses NFC. These features do work properly with OpenType and with TypeTuned fonts.
Bridging diacritics feature
The "Bridging diacritics" feature may have some unexpected behaviors when the feature is turned on with SIL TypeTuner. This feature is documented as Graphite-only and should not be expected to work properly in uniscribe applications. The feature works fine in Graphite applications although it does not work properly in conjunction with the Small capitals feature.
Wrong postscript names in version 4.106
Some characters were accidentally added to the fonts. Unfortunately they have the wrong postscript name and so if you choose to use them, there will be problems if you create a pdf. We anticipate removing these characters from future font builds. These are:
Small capital substitution in version 4.100
In InDesign and XeTeX (and probably any other application which uses true small capitals) the OpenType small capitals feature will not work properly, nor will any application-specific fallback mechanism. There are now new versions of Doulos SIL and Charis SIL which address this bug. Version 4.106 adds support for true small capitals.
U+0358 COMBINING DOT ABOVE RIGHT is not positioned correctly
This issue has been addressed in version 4.106.
Left-stemmed tone marks U+A712..U+A716 do not shape in Word 2003.
This is an issue with Word 2003. It has been reported to Microsoft. Even with a very new version of Uniscribe, Left-stem tone bars (U+A712..U+A716) may not ligate properly in Word 2003. We have found that if you type a Right-stem tone bar (U+02E5..U+02E9) immediately preceding the Left-stem tone bars they sometimes ligate. (However, that might not be desired in your document!) They do ligate properly in Notepad and OpenOffice if you have an appropriate version of Uniscribe (version 1.420.2600.2180 or later).
Combining marks do not shape properly over Cyrillic characters in Word 2007.
This issue was fixed in version 4.106 of our fonts.
WordArt has struggles with Unicode. Font linking may be going on, but the problem has not been clarified.
Some characters do not render properly in Internet Explorer
This is a problem in Internet Explorer for the following characters:
Microsoft has been alerted but we do not know a solution as yet.
The font does not support some combining marks
The following overlaid combining marks are present in the font but do not have attachment points and so will not render properly:
Cause: This is by design. For various technical reasons it is best to avoid using overlay combining marks. That is why, for instance, Unicode does not define a decomposition of U+026B LATIN SMALL LETTER L WITH MIDDLE TILDE to U+006C LATIN SMALL LETTER L + U+0334 COMBINING TILDE OVERLAY.
The following combining marks are not present in the font:
For reasons similar to the overlay diacritics, U+0321 COMBINING PALATALIZED HOOK BELOW and U+0322 COMBINING RETROFLEX HOOK BELOW are absent from the font ... by design. In this and the overlay diacritic case, Unicode provides for most uses of these marks through precomposed characters.
The Unicode standard deprecates U+0340 and U+0341, so we omitted those. The marks U+0342..U+0345 are primarily for Greek usage and, as mentioned above, the font is not intended to provide general support for Greek.
Combining mark sequences may be incorrectly rendered
This is not a bug in the font, but it is a Uniscribe bug which has been reported to Microsoft (update:it is reportedly fixed in Uniscribe version 1.0606.5078.0). It will only be a problem in applications using OpenType, not applications using Graphite. Failure depends on surrounding text. The following table lays out which character sequences will be a problem.
Base characters for which the indicated combining mark sequence may be incorrectly rendered.
Also, with versions of Uniscribe prior to Windows XP SP2 and Office 2003, the sequence U+006E LATIN SMALL LETTER N + U+0329 COMBINING VERTICAL LINE BELOW is incorrectly rendered as U+019E LATIN SMALL LETTER N WITH LONG RIGHT LEG
Combining marks in the U+1DC0..U+1DFF range
Combining marks in the U+1DC0..U+1DFF range may not position correctly unless you have an appropriate version of Uniscribe (version 1.626.6001.18000 seems to work). Even then, they may not position properly in Word 2003, Word 2007 or Word 2010. They position nicely in Notepad and OpenOffice.
Dotted digraphs with diacritics
Unicode specifies that, unlike i or j, etc., these characters do not lose their dots:
Mac OS X
Glyphs from version 4.106 are not displaying properly in Mac OS X (10.4 and 10.5)
This is a problem with version 4.106. We are investigating the cause of this issue. Currently the only solution is to revert back to version 4.104 (Charis SIL 4.104 and Doulos SIL 4.104) where they appear properly. Version 4.104 has an installation error, but it is safe to ignore that error.
Fonts do not appear in the main font menu under Mac OS 10.4+
In some applications the fonts do not appear in the main font list, they appear in the part of the font menu grouped with large fonts or non-Latin fonts. Under Word it may cause a switch to the Russian Phonetic keyboard and Cyrillic characters are inserted unless Word’s preferences are used to disable “match font with keyboard”. At the moment we do not know what is causing this behavior, but it is a known issue.
2012-03-21 SC: added KI about Firefox 11 and Graphite features
As this font is distributed at no cost, we are unable to provide a commercial level of personal technical support. We will, however, try to resolve problems that are reported to us.
Please note that this font is intended for use by experienced computer users. Installing and using this font is not a trivial matter. The most effective technical support is usually provided by an experienced computer user who can personally sit down with you at your computer to troubleshoot the problem.
Before requesting technical support, please:
If that fails to answer your question, or for more information, please contact us.