- From: John Hudson <tiro@tiro.com>
- Date: Fri, 30 Oct 2009 18:39:15 -0700
- To: John Cowan <cowan@ccil.org>
- CC: Peter Constable <petercon@microsoft.com>, Andrew Cunningham <lang.support@gmail.com>, "Martin J. Dürst" <duerst@it.aoyama.ac.jp>, LTRU Working Group <ltru@ietf.org>, www-font <www-font@w3.org>, Hċkon Wium Lie <howcome@opera.com>, www-style <www-style@w3.org>, Jonathan Kew <jonathan@jfkew.plus.com>, Stephen Zilles <szilles@adobe.com>, "Adam Twardoch (List)" <list.adam@twardoch.com>
John Cowan wrote: > Is the repertoire of characters considered part of the typographic > convention of a language? Because if so, hardly two languages will > share the same conventions. Not in the context of OpenType. The script and language system tags are hierarchical, so what the language system tags indicate are 'particular typographic conventions for the given script'. The classic case is Cyrillic <cyrl> Default <dflt> Serbian <SRB > The Serbian lookup set, in this case, would include mappings from default Cyrillic glyphs to preferred Serbian forms for certain letters, i.e. the Serbian typographic conventions for the Cyrillic script. As I wrote earlier, the OT language system tags are not necessarily linked to individual corresponding natural languages, but may relate to more than one natural language, and there are circumstances in which the users of one language might wish to invoke the typographic conventions of another. As I wrote in response to a query from John Daggett today regarding the latter situation: One scenario that comes immediately to mind, and I suspect Jonathan could suggest other examples from his experience at SIL, is when users of a minority language that is not supported explicitly in the OTL language system tags of a given font have cultural glyph preferences that are shared by language users whose language is supported in that font. The example I usually use is that of Macedonian Cyrillic, whose users may prefer the localised variant forms of some characters that are more widely recognised as associated with Serbian. There are a lot of fonts that have <SRB > language system support, but relatively few with explicit <MKD > support. This is because the Serbian preference has been quite well documented,* while the Macedonian preference has not. There are many languages and cultures for which the preferred typographic conventions are undocumented or, indeed, may be emerging in the case of languages recently reduced to writing. In such cases, there may be already-supported language systems in fonts that provide preferred forms, and enabling access to OT language system tags independent of other language tagging would enable users to access these forms without requiring updates to their fonts. Typographic conventions are not limited to language or typical locale preferences. Some conventions may be regional, affecting multiple languages within a geographic or market area that elsewhere use different conventions. In terms of the variant national typographic conventions of publishing of classical Greek texts in different parts of Europe, it makes sense even to speak in terms of Greek <grek> French <FRA > German <DEU > i.e. the French and German typographic conventions of use of the Greek script. I have done font development work for consortia of publishers and, although the situation has not yet occurred, I can imagine different publishers employing variant typographic conventions and, in the context of custom fonts, addressing these via custom language system tags. At this point, the connection of such tags to natural languages is completely severed and such a mechanism would only make sense if addressable independently of document language tagging. John Hudson * http://jankojs.tripod.com/SerbianCyr.htm
Received on Saturday, 31 October 2009 01:39:56 UTC