Johannes Ramminger at the Thesaurus Linguae Latinae Project (Munich) is gathering material for a Neo-Latin lexicon. He’s written (or had written for him) a program: you feed a Latin e-text into it, it identifies words not in the classical lexicon. Inadvertently, he’s come up with a text-correction program, it’s very like the one the Thesaurus Linguae Graecae Project here at Irvine uses to verify the accuracy of the texts it enters into its data bank. I wish Ramminger would make his program public, it would be useful as a spell-checker (except, of course, it would also identify postclassical spellings, such as foemina for faemina, so you’d have to weed these out to be left with genuine typos).

How on earth do we know this?  I asked this question to this group a few years ago and got no satisfactory answer.  I have a very simple need, which is a Latin spell-checker.  This would revolutionise our approach to digitisation of Latin texts. It is a relatively simple task, because I looked at the problem a few years ago.  I did not have the time or resource to solve it myself.  Perhaps I will look at the problem again.  But, then, there seems to be a whole humanities research industry devoted to 'digital humanities', so why hasn't someone provided a solution to this simple problem?

My pleasure: :)
http://drouizig.org/index.php?option=com_content&view=article&id=75&Itemid=68&lang=en

(disclaimer: this Latin language spellchecker is a very basic tool, intended to help proofreading gross errors in OCRized Latin texts, or typos in the ones we type up ourselves; it in in no way exhaustive or perfect, and it DOES NOT PRETEND TO DO a latinist's job regarding the grammar; all it does is pointing to you words that are likely to be mistakes - and naturally, it's more useful on literary texts than on, say, charters with inventive spelling. Yet... it exists, and it seems to do many people a lot of services).