1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
|
Attributing inscriptions to masons on the basis of their style is
useful for dating and reconstructing texts. A team of computer
scientists, in collaboration with the epigrapher Stephen Tracy, has
developed digital methods to quantify letter shapes, which can then be
compared statistically across texts to determine likely
authorship. Promisingly, the results have so far agreed entirely with
Tracy’s own attributions. Digital stylometry, however, must not be
fetishized as a fountain of “essentially more objective,” “hard
evidence” or of “correct and unambiguous” facts (Papaodysseus et
al. 2007: 749; Tracy and Papaodysseus 2009: 101; Panagopoulos et
al. 2009: 1404; also see Tracy 2003: xviii). Computerization does
eliminate the inconsistencies of human perception and enable great
precision, but any promises of scientific objectivity should be
scrutinized. This paper attempts to provide such scrutiny by advancing
three sets of claims. First, epigraphy possesses a troubled history
with supposedly objective criteria: the now-discredited
three-bar-sigma dating rule, for example, held the field in thrall for
much of the twentieth centurybefore being disproved (Chambers et
al. 1990); the episode’s clear moral is that we should be skeptical
about other, similar claims to objectivity. Second, comparison with
the connoisseurship of Athenian pottery underscores the relative
poverty of epigraphical connoisseurship: Tracy is the only recent
Hellenist to devote sustained effort to identifying epigraphic hands,
resulting, inter alia, in no robust scholarly consensus on the
criteria or standards of judgment. Finally—and most important—major
epistemological issues simply remain unconsidered. Essentially,
Tracy’s method has been refined into a set of algorithms. But the fact
that analysis takes place in a computer does not guarantee that the
method is correct, nor does it make its results “objective” in any
deep sense. The method and its calibration remain exercises in human
judgment, grounded ultimately in our visual experiences. Digital
imaging calls for reinvigorated, updated engagement with conceptions
of style and connoisseurship, not their repression.
|