We are developing data visualis/zation tools to explore sets of versions/rewritings/retranslations -- of ANY work.



  • Rafts of Translations - British Centre for Literary Translation (October 2017) and University of Amsterdam (2017) - slides at foot of this page

  • BBELT presentation (Mexico 2016) slides attached below (foot of page)

  • VVV presentation at Moscow University of the Humanities, and discussion with Boris Gaydin, November 2015. Video. Report.

  • CODAH 2015 slides attached below.

  • TEDx Swansea presentation on VVV by Tom Cheesman (June 2015) - slides attached below

  • VVV presentation by Cheesman, Flanagan, Thiel at Digital Humanities 2012 (Hamburg) - abstract - video (28 mins)

  • Slides 2013 including some very old ones!

Other presentations of our work during 2012-13:

Our prototype map of Othello translations was also reviewed in Wired

Papers attached at the foot of this page:

Items below on this page by TC:

  • Stylometric analysis of 40 German versions of Othello - with Jan Rybicki

  • Alignment Maps of versions of Othello designed by Studio Nand

  • Tag Line for 'Her father loved me...' by Zhao Geng

    • 'Far More Fair Than Black' - French and German Translators' Choices

  • Table of versions of 'Virtue? A fig!'

  • The Eddy Variation Overview Chart in the VVV system: identifying where one translator is most unique

  • Two-line and equivalent segments (60-100 characters) in Othello 1.3, with speakers and poetic form, in order of 'Viv'

Stylometric analysis of 40 German versions of Othello

Tom Cheesman and Jan Rybicki, July 2014

The visualization below represents a stylometric analysis of 40 German versions of Othello. Dr Jan Rybicki carried out the analysis which generated the graphic.

Comments by TC below the image. Click to enlarge it.

The analysis is based on part parallel samples, each of approximately 8,000 words (from the opening scene of the play, forwards). Text samples based on OCR output were copy-edited to 96% or better accuracy.

Rybicki used the 'stylo' package for R.* This programme compares frequencies of Most Frequent Words in each version, using multiple different analytic parameters (different cullings of word classes, different ranges of frequencies, etc), and compares the results of these many different analyses, to produce an aggregate analysis, which is visualised using GEPHI tools. *See: M. Eder, M. Kestemont, J. Rybicki, 'Stylometry with R: A Suite of Tools', Digital Humanities 2013: Conference Abstracts, pp.487-89; accessible here.

Labels show the (main) translator's surname and the date of publication or creation. The suffix "_T" indicates a theatre script, i.e. a version which has not been published in book form. The suffix "_Pr" (prose) indicates a version for study reading. The prefix "Baud" indicates a version of (or based on) the translation by Wolf Baudissin (1832, in the Schlegel-Tieck edition).

Spatial clusters indicate general stylometric similarities among versions, as contrasted with others.

Connecting lines indicate specific similarities found by some set of individual analyses; the thickness of lines indicates the strength of these similarities across multiple analyses. -- For example, the line at top centre, connecting Gildemeister (1871) and Vischer (1887), shows that those two texts have a lot of stylometric features in common. A little below them, Bodenstedt (1867) connects along a very thick line to Ruediger (1963). In fact, Ruediger rewrote Bodenstedt's version. Far below left, Rothe (1956) is distantly, thinly connected to Engler (1976). But it is very unlikely that this line represents any real intertextual relation, in the sense of a genealogical link.

The corpus includes several sets of versions identified as clusters by the analysis. Individual anomalies and connecting lines are also of interest. Some observations:

  • All the 4 versions with suffix _Pr, annotated in GREEN, form a clear cluster (bottom right). They differ from all the rest in terms of genre: these versions were created for study reading, rather than for leisure reading, or stage performance. Two of these are from the 18th century, two are from the later 20th century. Thick connecting lines may imply intertextual links. Close mutual proximity and distance from the rest of the corpus, indicate that any differences found in the rest of the corpus cannot be due to historical changes in the German language. They must be due to historical changes in German literary culture.

  • All the 4 versions from the early 19th century (except one: Baudissin, 1832 etc) are in a distinct BLUE cluster (centre right). This cluster and others show the importance of period style.

  • All the 4 versions from the later 19th century, in PURPLE, form a weak spatial cluster (top centre), but do appear in the same graphic space, relative to all others. This again indicates a period style.

  • 9 versions shown in ORANGE, each labelled "Baud--", are based on the translation by Baudissin (first in the Schlegel-Tieck edition, 1832). Note that the distances between these versions in the graphic exaggerate their stylometric differences. On the other hand, this is a much-redacted, 'canonical' text.

    • Baudissin versions furthest from the centre are, as might be expected, two theatrical adaptations of it (Engel 1939, upper left; Lauterbach 1972, lower centre) -- and Wolff's (1926, lower right) print-published version. That is a more radical rewriting of Baudissin's text than its title page suggests.

    • Thick lines connect all versions within the Baudissin group, and connect beyond it: with the blue and purple clusters, as well as with several uncoloured versions: reading clockwise from top centre, Schroeder (1962), Zeynek (1948), Schaller (1959), Gundolf (1909), Flatter (1952) and Schwarz (1941) -- these all appear to fall within the 'orbit' of the Baudissin text.

    • Bab and Levy's redaction of 1923 appears most central, but that is an accident. We could analyse dozens more versions of the Baudissin text. Also, we do not yet have a digitized version of any of the first printings.

  • All the 4 versions from the 21st century form a very clear period style cluster, in YELLOW. These are all theatre script versions (Zaimoglu's version was published, but it was initially a theatrical commission). Each one adapts (rewrites, transforms) the play in a very different way - but all are using language in much the same way, or more precisely: in a similar way relative to all previous translations.

  • A cluster without colour annotation is evident at lower left. This includes several versions of the 1970s to 1990s, as well as (lowest left) Rothe's version from the 1950s. It would seem that Rothe was far ahead of his time.

  • Schroeder (1962), Ruediger (1983), Zeynek (1948) are squarely in the '19th-century space': archaicizing versions.

  • Fried (1972) interestingly occupies a stylistic space of his own, midway between the 19th- and 21st-century cluster spaces. If we can read the graphic narratively, then he is exploring a stylistic space which late 20th- and 21st- century translators will occupy.

  • Schwarz (1941), a forgotten translator - and the only woman translator here - is interestingly close to Fried: ahead of her time?

  • Guenther (1992, lower left) is currently the best-known German Shakespeare translator. His version shows numerous links both to precursors and to later versions. Lines connect Guenther back in time to Fried (1972), and also to Laube (1978), who in turn connects back to Swanczynna (1972), who in turn connects back to Gundolf (1909) -- a 'genealogy' of translation styles?

  • Guenther's version is known to be highly influential. It connects forwards in time to Buhss (1996) (which in turn connects forward to Zaimoglu 2003, which connects very strongly with Karbus 2006), and also to Wachsmann (2005) and Leonard (2010).

  • But we cannot tell from such analysis whether stylistic similarities are due to conscious or unconscious borrowing or imitation (intertextual geneaology), or due to sharing a period-specific style. A line links Guenther (1992) and Motschach (1992) (Motschach is a prolific but obscure Shakespeare translator). Neither is likely to have been able to consult the other's simultaneous translation.

Alignment Maps of Versions of Othello by Studio Nand

Click to enlarge the image.

Each pair of 'bar-codes' shows: on the left, the speech-structure of Shakespeare's Othello, Act 1 Scene 3, and on the right, the speech structure of a German version (translation or adaptation), labelled with writer and date.

Thinner, linking lines show alignments.

In each 'bar-code', horizontal bars represent speeches. Their thickness represents the length of the speech in words.

The speech highlighted in red is Othello's speech 'Her father loved me...'

We can see at a glance where versions expand, contract, omit from, add to, or re-order material from the English play text.

These alignment maps form the basis of navigation in the 'Parallel Text' and 'Alignment' views at www.delightedbeauty.org/vvvclosed.

Tag-line by Zhao Geng (5 April 2011) using Chirag Mehta's © algorithm and Erik Arvidsson's © dynamic slider control. This visualisation displays the least frequent words in 22 versions of Othello's long speech to the Senate about his life-story (Act 1, scene 3).

'Far More Fair Than Black' - French and German Translators' Choices

The table below shows choices for 'fair' and 'black' made by 55 translators of the Duke's last couplet: 33 in German and 22 in French. (Go to 'Materials' to see all the texts.) Differences which might be significant - statistically and culturally - are highlighted in red.

I divided the translators into 'famous' and 'obscure' translators. (This was for a presentation at the 2010 Author-Translator conference.) The subset of the 'famous' includes (a) writers whose names have a special cultural standing, and (b) celebrated rewriters of (a lot of) Shakespeare's work. The 'famous' are: in German, Wieland, J.H.Voss (3 rewritings), Schiller, Baudissin, Gundolf [George], Rothe, Fried, Günther, Zaimoglu/Senkel; and in French, Le Tourneur (2 rewritings), de Vigny (2 rewritings), Hugo (2 rewritings), Aicard, Robin, Jouve, Bonnefoy, Déprats. So that's 11 rewritings by 'famous' translators in each language

Q: What is different about French and German choices? What might explain these differences? Is it that the languages have different expressive resources? Or is it the cultures rather than the languages? Ideological constraints? Is it the specific cultural traditions of Shakespeare translation?

Q: Is there anything different about the work of rewriters whose name has a special cultural standing? What might explain the differences between French and German, as regards the relation between the work of the 'famous' rewriters and the others?

Q: Would we find similar patterns if we investigated a different speech?


April 2016

Here is a table of versions of 'Virtue? A fig!' - click to enlarge

April 2016

Hedwig Schwarz (1898-1985) translated 20 of Shakespeare's plays between 1936 and 1947.

Using the Eddy Variation Overview Chart in the VVV system, we can see,

first image below, how variation maps across ALL versions of Othello 1.3, and

second image below, where Hedwig Schwarz's translation differs most (and least) from others.

NB: Text segments of versions are displayed when the relevant node is brushed - one at a time, in the live system.

April 2016

From paper for DSH.

Click to enlarge.

The table shows two-line and equivalent segments (60-100 characters) in Othello 1.3, with speakers and poetic form, in order of 'Viv': quantified variation in variation, based on 20 German translations.

Christoph Martin Wieland - with Goethe, Schiller and Herder

Wieland's Shakespear Theatralische Werke, vol 8 of 8, 1766 (Taylor Institute copy at Google Books)

Page images from the Taylor copy of Vol 1 with King Lear

Translation of Wieland's footnote p. 305:

Here the original is more daring than the [my] translation. Shakespeare has Regan ask: have you never found my brother's way To the forfended place?

Translation of Wieland's footnote, p. 320:

This nonsensical babble has been allowed to stand almost as confused as it is in the original, as a sample of a very common vice of Shakespeare's: that of only half expressing his thoughts, hurling ill-matched metaphors together, and dispensing himself from all the rules of grammar.

This refers to Edmund's speech in King Lear 5, 3.:

In wisdom I should ask thy name;

But since thy outside looks so fair and warlike,

And that thy tongue some say of breeding breathes,

What safe and nicely I might well delay

By rule of knighthood, I disdain and spurn.

Back do I toss those treasons to thy head;

With the hell-hated lie o'erwhelm thy heart;

Which - for they yet glance by and scarcely bruise -

This sword of mine shall give them instant way

Where they shall rest for ever. Trumpets, speak!

Pages of the revised, pirated edition by Gabriel Eckert (vol 2 of 22 vols, 1778-83) of Eschenburg's Shakespeare [first edition 12 vols 1775-77 + 1 vol 1782]: from 'The Two Gentlemen of Verona'