[Humanist] 24.138 digital palaeography

Humanist Discussion Group willard.mccarty at mccarty.org.uk
Wed Jun 23 07:12:12 CEST 2010


                 Humanist Discussion Group, Vol. 24, No. 138.
         Centre for Computing in the Humanities, King's College London
                       www.digitalhumanities.org/humanist
                Submit to: humanist at lists.digitalhumanities.org

  [1]   From:    Peter Stokes <peter.stokes at kcl.ac.uk>                    (102)
        Subject: Re: [Humanist] 24.136 digital palaeography?

  [2]   From:    "Desiree Scholten" <desiree_scholten at hotmail.com>         (21)
        Subject: Re: [Humanist] 24.136 digital palaeography?

  [3]   From:    John Unsworth <unsworth at illinois.edu>                     (98)
        Subject: Re: [Humanist] 24.136 digital palaeography?

  [4]   From:    Robert Kraft <kraft at sas.upenn.edu>                       (111)
        Subject: Re: [Humanist] 24.136 digital palaeography?


--[1]------------------------------------------------------------------------
        Date: Tue, 22 Jun 2010 10:55:27 +0100
        From: Peter Stokes <peter.stokes at kcl.ac.uk>
        Subject: Re: [Humanist] 24.136 digital palaeography?
        In-Reply-To: <20100622085343.8743C59F84 at woodward.joyent.us>

Dear Willard (and list),

I think a lot of the questions you raise apply not just to 'digital' palaeography but to palaeography in general, and this is one aspect of the 'crisis'. There has been an ongoing debate in the field over whether palaeography is (or should be) an 'art' or a 'science', and this debate began well before digital methods began to be applied (see, in particular, articles in Scrittura e Civiltà from 1995 and 1996, but the question goes back to work done by L. Gilissen in the 1970s).

The question of letters as 'embodied in time' has also been addressed by 'traditional' palaeographers, for want of a better term, and different people have naturally taken different approaches. Albert Derolez, for one, defended his use of 'morphology' over 'ductus', in his recent book on the palaeography of Gothic script. Ultimately, though, like with everything else we do, I think it depends on what question you want to answer. Scribal attribution, like authorship attribution, is arguably amenable to quantitive analysis and therefore computing, and this is where most of the work has been done. I have also argued that scribal attribution, like authorship attribution, has the fundamental problem that we lack ways of verifying or even debating the results. Forensic document analysts have shown that examining 'strokes in time' is certainly useful for writer identification, but that capturing this from a finished product is very difficult; they have certainly tried to model the mechanics of human movement, though. The historical development of scripts, however, depends almost entirely on analysis of strokes in time (or ductus), as Mallon demonstrated, and so this is much less amenable to digital analysis. Categorisation of scripts is more complex: computers are good at categorising, in one sense, but their categories may not match the scholars' which raises interesting questions about what we do with those results. On the other hand, simply trying to teach a computer to identify 'Caroline minuscule' raises further useful questions about what we as scholars mean by Caroline minuscule in the first place, and indeed some palaeographers have argued that any categorisation of script is an artificial and essentially meaningless exercise anyway.

I've discussed much of this in my article 'Digital Palaeography: Present and Future', and I'd suggest that the other contributions to that volume give a pretty good sense of the 'state of the art', although I think all of us in that book are guilty of downplaying the question of movements in time. The whole book is available at <http://www.i-d-e.de/schriften-2/kodikologie-und-palaographie-im-digitalen-zeitalter>. There are also two articles in the Digital Medievalist journal on automatic scribal identification and classification:  http://www.digitalmedievalist.org/journal/1.1/ciula/  and  http://www.digitalmedievalist.org/journal/3/stokes/ , as well as various other works referenced in those.

Yours,

Peter

--
Dr Peter Stokes
Research Associate (Analyst)
Centre for Computing in Humanities
King's College London
Room 210, 2nd Floor
26-29 Drury Lane
London, WC2B 5RL
Tel: +44 (0)20 7848 2813
Fax: +44 (0)20 7848 2980

On 22 Jun 2010, at 09:53, Humanist Discussion Group wrote:

>                 Humanist Discussion Group, Vol. 24, No. 136.
>         Centre for Computing in the Humanities, King's College London
>                       www.digitalhumanities.org/humanist
>                Submit to: humanist at lists.digitalhumanities.org
> 
> 
> 
>        Date: Tue, 22 Jun 2010 09:40:48 +0100
>        From: Willard McCarty <willard.mccarty at mccarty.org.uk>
>        Subject: digital palaeography?
> 
> 
> I would like to get a grip on the current state of thinking about digital
> palaeography. I'd like to know in broad terms, techno-intellectually, (a)
> what now is demonstrably possible; (b) what we have good reason to think we
> can do in the relatively near future; and (c) where we can dream of going
> someday. Provision of what we now call "high-definition" images is a given.
> That's fine. But that's not the end of the matter, I hope.
> 
> Surely with palaeography as elsewhere throughout the humanities, there is a
> fundamental, enlightening but unavoidable difference between the precision
> of digital tools and the rather different qualities of scholarship. Since
> Turing's definition of computability and the machines that have followed, we
> have tended to go around and around in a vicious circle, from that which can
> be precisely stated to that which can be computed and back again. Which
> leaves anyone not in that circle imprecise, vague, muddled....
> 
> Some write about a "crisis" in palaeography, by which I gather they mean to
> say that the intermixture of human judgement (which cannot be computed) and
> historical fact (which presumably can) is a problem, the kind that can be
> solved -- or at least that having much less of the former and more of the
> latter is a good and achievable thing. Digital tools are supposed to help.
> But with what, exactly, can they help the palaeographer, without the
> pretense that handwritten letterforms are a different kind of thing than
> they actually are?
> 
> Is the root crisis the problem of being an embodied creature in time?
> 
> Of course we can choose to ignore the fact that *every* letterform is a work
> of art, however highly repetitive its production, however far to the back we
> shove its individual aesthetics. (Look at the Book of Kells in Dublin and
> then tell me that those aesthetics aren't important.) Every letterform is
> something produced by a living hand at a particular historical moment in a
> particular place, though one or more of these coordinates may be unknown. Of
> course we want to make more of the unknown known (to "push back the fence of
> the law", as Jacob Bronowski said), but great scholarship needs unblinking
> recognition of what we don't, and perhaps can never, know. Yes?
> 
> And, it seems to me, there's a persistent tendency to downplay the fact --
> stressed by calligraphers, who like painters to the art historian tend to be
> suspect because they *do* what the scholars study -- that handwritten
> letterforms are traces of movement in time involving moment-by-moment
> variation in the conditions of production. The workaday scribe (as opposed,
> say, to a Jackson Pollock of the edged pen) will want to reduce that
> variation to an absolute minimum, but part of being skilled involves the
> ability to respond instantaneously to the unexpected, e.g. a defect of the
> writing surface. Logically it would make sense, I'd suppose, that capturing
> the dynamics of production would get us closer to the identity of the
> historical writing hand than mere classification of, say, a Carolingian
> minuscule "a" with such-and-such a characteristic bowl or serif. And it just
> might open up new questions.
> 
> Is progress being made on that front?
> 
> What is the state of the art? (Note I didn't say, the state of the science,
> but that, too, is a question and needs to be asked simultaneously.)
> 
> Comments?
> 
> Yours,
> WM
> --
> Willard McCarty, Professor of Humanities Computing,
> King's College London, staff.cch.kcl.ac.uk/~wmccarty/;
> Editor, Humanist, www.digitalhumanities.org/humanist;
> Interdisciplinary Science Reviews, www.isr-journal.org.


--[2]------------------------------------------------------------------------
        Date: Tue, 22 Jun 2010 14:27:19 +0200
        From: "Desiree Scholten" <desiree_scholten at hotmail.com>
        Subject: Re: [Humanist] 24.136 digital palaeography?
        In-Reply-To: <20100622085343.8743C59F84 at woodward.joyent.us>

I generally hold my quiet on discussion lists as I am rather fresh in the 
market so to speak (will begin my PhD in October) but last year I looked a 
little deeper into this matter myself. On my academia.edu page you can find 
a paper I wrote last year on the state of the art in digital paleography. My 
conclusions were that technology can bring us much further when it comes to 
enhancing images. This is particularly helpful when working with 
palimpsests, or damaged manuscripts. Still, computers lack the creativity of 
the human mind, which is needed when it comes to recognizing letterforms, or 
classifying a script by its general appearance. Everything which leans on 
hard to define standards needs such interpretative creativity and 
flexibility, which cannot be programmed. Also, my experience thus far with 
paleography has been that experience is very important, a good paleographer 
knows the basic traits of a script, but has seen so many individual examples 
of that script that he or she is able to place it in a broader perspective. 
Again, this is not only hard to teach to a student, but even harder to teach 
to a program which can work with more or less rigid definitions only.

Kindest regards,
Desiree Scholten


--[3]------------------------------------------------------------------------
        Date: Tue, 22 Jun 2010 08:50:56 -0500
        From: John Unsworth <unsworth at illinois.edu>
        Subject: Re: [Humanist] 24.136 digital palaeography?
        In-Reply-To: <20100622085343.8743C59F84 at woodward.joyent.us>

Willard,

I was recently at the MARGOT conference on the Digital Middle Ages, at Barnard, where this plenary panel occurred:

 Plenary Panel:

Quantitative Palaeography through Massive Image Analysis: The Graphem Project

 Chair: Denis Muzerelle, IRHT, CNRS, Paris

 -       Marc H. Smith, École des Chartes, Paris, and Maria Gurrado, IRHT/CNRS, Paris

"Beyond Typology: Rethinking Palaeographical Categories with Computer Science?"-       Hubert Empotz, LIRIS-INSA, Lyon and Mathieu Exbrayat, LIFO, Orléans

"New Tools for Exploring, Analysing and Categorising Medieval Scripts"

-       Dominique Poirel, IRHT/CNRS, Paris

"Access to Textual Contents of Medieval Manuscripts Using Wordspotting Methods"

I don't know if any of these folks subscribe to Humanist, but perhaps if they do, they'd be willing to share presentations from this panel.  

John Unsworth



--[4]------------------------------------------------------------------------
        Date: Tue, 22 Jun 2010 20:58:45 -0400
        From: Robert Kraft <kraft at sas.upenn.edu>
        Subject: Re: [Humanist] 24.136 digital palaeography?
        In-Reply-To: <20100622085343.8743C59F84 at woodward.joyent.us>

Ah yes, digital paleography. It has been more than 20 years since I 
tried to sketch out what I needed for working with the thousands of 
papyri and related fragments in the UPenn collections (primarily the 
Greek and Coptic), but have not been able to find a programmer to take 
up the challenge, although I've been told that it "should be possible." 
Here are some steps that seem to be relatively straightforward:

(1) Establish the base line of writing;
(2) Judge the degree of tilt (if any) to the writing;
(3) Measure the average height of letters;
(4) Judge the width of individual letters, where possible (easiest for 
uncials);
(5) Evaluate color of writing surface and of the ink;
(6) Identify individual letters by shape (OCR algorithms);
(7) Isolate unusual shapes for closer attention.

My specific needs include matching fragments that may be from the same 
document, and such "digital paleography" would be a long step in that 
direction. Further steps could include pattern matching (edges and fiber 
patterns) of fragments identified as containing similar writing, but at 
that stage, close visual attention would also be very important.

If some program(s) along these lines exist(s), I'd be anxious to put 
it/them to use. Digital images of the fragments are available, and I'm 
about to spend time attempting to group them paleographically.

Bob Kraft, UPenn Emeritus



More information about the Humanist mailing list