Twelve Hamlets: A Stylometric Analysis of Major Characters' Idiolects in Three English Versions and Nine Translations

Following my own comparative stylometric analyses of
originals and translations of literary works (Rybicki 2006,
2006a), I have decided to expand the scope of this research to
more than just two languages. This has been made possible by
the fact that Burrows’s well-established method, first used in
his study of Jane Austen (1987), and later developed, evaluated
and applied by a number of scholars, including Hoover (2002),
can be applied to the most frequent words of any text, also in
a language unknown (or less-known) to the researcher. Quite
self-evidently, Shakespeare’s Hamlet was chosen for its status
of a crucial work of English literature and its numerous
translations. The fact that the English Hamlet exists in three
primary versions (First Quarto, Second Quarto, First Folio) was
also a serious incentive.
Character parts were extracted from the three above-mentioned
originals and nine translations: Czech (by Josef Jirí Kolár,
1855), French (François-Victor Hugo, 1863), German (August
Wilhelm Schlegel, 1798), Hungarian (János Arany, 1864),
Italian (Goffredo Raponi, 1999), Polish (ca. 1875), Portuguese
(1966) Russian (Mikhail Morozov, 1954), and Spanish (Leandro
Fernandez de Moratin, 1798). The selection of the translations
was a compromise between their age (the older the better) and
availability in electronic form. The three originals were taken
from the collection of exemplary electronic texts on TACT CD
(1996). Most frequent words were identified in each version,
basing on the rule that a given word was allowed if it appeared
in at least 5 of the major character parts: these included
Claudius, Gertrude, Hamlet, Horatio, Laertes, Ophelia and
Polonius. As a result, the number of the most frequent words
included in the analysis varied from 157 (First Quarto,
obviously the shortest text) to 282 (the Italian translation).
Relative frequencies were then obtained for each character’s
use of each of the above-mentioned words; matrices of the
relative frequencies were then used to produce multidimensional
scaling (MDS) graphs for each version of the play. The same
material was also treated with cluster analysis, usually with
good agreement with MDS.
The following observations have been made:
1. Graphs for each version of the play always included at least
two peripheral data points, invariably including the two
female characters, Gertrude and Ophelia;
2. Laertes and Horatio frequently joined Gertrude and Ophelia
in their peripheral orbits around the data point for Hamlet,
usually central in all graphs (with the notable exception of
the Czech translation);
3. In many translations, the idiolect of Hamlet was more or
less similar to those of some other characters: its closeness
to Polonius (or Corambis) in the First Quarto lessened in
the Second, only to be replaced by a much stronger
similarity to Laertes. An even stronger similitude was
observed between Hamlet, Polonius and Claudius in the
Portuguese translation, or to Claudius alone in the Italian
version. In general, Polonius and Claudius were the idiolects
most consistently similar to that of the main character.
4. The closest similarity of pattern between entire graphs was
that between the First Quarto and Schlegel’s German
translation (Hamlet and Polonius surrounded by even-spaced
peripheral characters);
5. The general pattern was also quite similar between the
Second Quarto and the Folio (peripheral Gertrudes, Horatios
and Ophelias, central Hamlets, other characters in-between);
it was also roughly reproduced in the translations by Arany,
Moratin and Paszkowski.
6. The clearest gender division was observed in the Czech,
Hungarian, and Russian translations.
This introductory study seems to offer several promising
avenues for development. The similarity of pattern between the
Second Quarto and the Folio is not surprising; even less so is
its above-mentioned reproduction by three translations, since
their source material was usually a compilation of the two
original versions. On the other hand, Schlegel’s similarity to
the First Quarto seems pure coincidence, as the so-called “bad
quarto” was discovered more than twenty years after the early
German translation. The peripheral position of female characters
in all graphs is perhaps the most consistent feature of most of
European writing so far analysed with MDS; this effect also
travels very well in translation.
Yet the greatest potential for expanding this study lies in the
sheer number of Hamlet translations. In the languages included
in this study, they are at least five, and very often more than
ten; cultures such as Polish, German, or French, have produced
almost twenty Hamlets each. The inclusion of at least some of
these would provide a fuller picture of patterns of stylistic
differences between Shakespeare’s fundamental play and its
various realisations in other languages. Bibliography
Burrows, J. F. Computation into Criticism: A Study of Jane
Austen’s Novels and an Experiment in Method. Oxford:
Clarendon Press, 1987.
Hoover, David. "New Directions in Statistical Stylistics and
Authorship Attribution." ALLC/ACH 2002 Conference
Abstracts. Tübingen: Tübingen University, 2002.
Rybicki, Jan. "Burrowing into Translation: Character Idiolects
in Henryk Sienkiewicz's Trilogy and its Two English
Translations." Literary & Linguistic Computing 21.1 (2006):
91-103.
Rybicki, Jan. "Can I Write like John le Carré?" Digital
Humanities 2006 Conference Abstracts. Paris: CATI, Université
Paris-Sorbonne, 2006a.