Some more information and examples here: http://ajami.wikispaces.com/

Does Unicode currently cover all the modified letters needed for representing these Ajami texts?

Yes, I believe so. Anything missing at this stage would be rare and/or poorly documented.

Are there widely available Arabic fonts that are suitable for representing Ajami texts for scholarly or everyday purposes?

The Adobe Arabic fonts, which ship with Acrobat 7.0.5 and later and are available for purchase from the Adobe type site, should cover all the necessary characters. The MS Arabic Typesetting font might also, although I think some characters have been added to Unicode since this font was last updated.

The Tasmeem plug-in fonts for InDesign ME should also work for all these languages.

Note that in all these cases, I'm talking about the suitability of fonts from a technical perspective, i.e. Unicode characters support and appropriate layout.

I have a feeling that since there are several different Ajami systems each with different modifications to the basic Arabic letters, current Arabic fonts would be inadequate.

A lot of Arabic fonts support only the Arabic, Persian (Farsi) and Urdu character sets, since these represent the minimum set defined by the Windows 8-bit charset for Arabic. You need to look for fonts that support ‘extended Arabic’ and target the entire Unicode basic Arabic block and extensions.

What would be the ideal typographic solution for Ajami? What style of script would be preferable based on these samples?

The texts shown on the West African Islam website have been presumably written out by the scholars, so represent fairly characteristic but undistinguished handwriting. You see that the style is mostly quite flat, a descendant of the maghribi ‘cursive kufic’ style. In an earlier thread on the Arabic forum, I posted some images of sub-Saharan calligraphy, which are representative of the African style, commonly referred to as ifriqi. I like this style very much, but am not aware of any fonts that seek to represent it. [Titus Nemeth recently made a very nice revival of a metal maghribi type, but it is in a North African style which may or may not be appropriate for sub-Saharan languages.]

Thank you, John, for the detailed and helpful answer.

A lot of Arabic fonts support only the Arabic, Persian (Farsi) and Urdu character sets, since these represent the minimum set defined by the Windows 8-bit charset for Arabic.

Yes, that's what I was afraid of. But good to know that Adobe Arabic fonts support the whole Unicode Arabic block and extensions.

At the link in my second post, there is a table (PDF) of modified letters used in Wolofal, the Ajami script for Wolof (you will notice it's been assembled from several fonts). Several of the letters seem to be those already encoded in Unicode, like the ain with three dots above (U+06A0) that stands for /ŋ/. Others look like they should be encodable with combining marks, but I'm not sure how the very frequent three-dot combination is supposed to be encoded. Surely not by U+06DB, "Arabic small high three dots" found among the Koranic annotation signs section?

Then again, scholarly interest in Ajami began only very recently, so letters needed for representing Ajami may still be being added to Unicode.

Thanks for reminding me of that sample of ifriqi calligraphy. It would certainly be great to have a font seeking to represent that style. But realistically, a standard Naskh font seems most likely to be used in a typographic transcription of Ajami. At this point, though, I don't think Ajami has been represented in type very often.

If I knew anything about Arabic typography and creating Arabic fonts, I think creating a font suitable for representing the various Ajami scripts would be an interesting and rewarding project to pursue.

I have only now discovered that the Unicode Technical Committee has accepted Arabic Pedagogical Symbols for inclusion in Unicode 6.0, which is scheduled for release later this year. Here are a couple of the proposal documents:

http://std.dkuug.dk/jtc1/sc2/wg2/docs/n3460.pdf
http://std.dkuug.dk/JTC1/SC2/wg2/docs/n3460-A.pdf

These would add the Nuqta (the dots to indicate phonetic modification) in various numbers and positions as combining characters to the basic shapes of Arabic, and at first glance would allow the encoding of all the modified letters used for Wolof Ajami shown in the table (sorry, the link in my earlier post is wrong).

This seems a broader and more elegant solution than encoding each combination of basic shape and dots as pre-composed shapes. In fact, I was surprised it hadn't already been implemented, because I was searching for these combining symbols in vain in the current Unicode charts. Considering that extensions to the Arabic script have made use of such combinations for countless languages and dialects and that these orthographies have been ever changing, it is probably hopeless to document every single combination that ever occurred.

Perhaps there are other combining symbols used for other Ajami variants that have yet to be documented and added to Unicode, but it seems likely that by the end of this year we should be able to represent at least the most prominent examples of Ajami in Unicode.

Brian, the nuqta dot ‘symbol’ characters are currently making their way through ISO 10646 national balloting processes, and are expected to be added to one of the Arabic Presentation Form blocks. The final character names are different from those in the proposal documents, and they are being referred to as ‘Arabic Symbol’ characters.

Note that these are not combining marks: they are spacing symbol characters for use in e.g. pedagogical materials about the Arabic script. They are expressly not to be used for generative purposes: Unicode's Arabic encoding model remains letter-centric, not archigraphemic, and there are no combining dots that can be encoded with dotless archigraphemes to produce new letters. One can argue, as I would, that an archigraphemic encoding would have been more elegant and flexible, but that is not the kind of encoding that Unicode inherited from earlier standards, and the inclusion of these nuqta symbol characters does not open the door to such an encoding. Trying to have both a letter and archigraphemic encoding would require new canonical decompositions for existing letter characters, which is contrary to stability agreements between Unicode and other standards.

I see. One of the proposals made it sound like these would be used like combining marks. So they are only going to be spacing, non-combining symbols, and as such won't be able to be used to form Ajami letters not otherwise encoded in Unicode.

This is rather disappointing. I see why Unicode tries to preserve compatibility with earlier standards, but in some cases I wish we could just start over from a clean slate.

I suppose we couldn't allow archigraphemic encoding just for representing unencoded letters as long as it was prohibited to use them for composing already encoded letters? Yeah, I guess that would just complicate the encoding model for Arabic. And I suppose adding every possible combination of base letters and dots would be out of the question...

It looks as though more Ajami will be made available as of Unicode 6.1, in the Extended Arabic-A section of the BMP, between U+08A0 and U+08FF:

http://std.dkuug.dk/jtc1/sc2/wg2/docs/n3882.pdf

More manuscripts have gone online with Boston University's African Ajami Digital Library:

http://dcommon.bu.edu/xmlui/handle/2144/2381/browse?type=title

That certainly is welcome news that Ajami is getting enough attention that a Unicode inclusion proposal is being made. Thanks for sharing it as well as the Boston University online resources, Charles.

I don't know if there is a connection with this, but images I recently saw on the Web of manuscripts from a large ancient library found in Timbuktu appeared as if they might have been written in a script of this class.

However, looking this up on Google led me to an exhibition of documents from that library which were not merely written in Arabic script, but which were in the Arabic language, not in Bambara or anything else.