Analysis of Writer-Text-Translator Social Networks

1. Introduction

This paper presents an attempt to help understand the complex relationships between writers, their texts and their translators. A text by a writer may be translated by a single translator, or rival translations of the same text may appear; or a writer’s one text may be translated by one translator, and another of his texts by another translator; all this creates a complex mesh of connections that begs to be analyzed as a social network. The fact that some translators limit themselves to just one source language and one target language – the usual situation – does not mean that exceptions do not exist: Polish-Jewish anarchist poet Joseph Bovshover translated Polish into English (Sienkiewicz) and English into Yiddish (Shakespeare); Polish translator Ireneusz Kania translates (directly!) from 20 different languages (including Sanksrit and Tibetan). One of the main aims of this reconnaissance was to establish the degree of clustering by language, i.e. whether single-source-language communities are also interconnected.

2. Material and methods

Data on writers and translators was obtained from UNESCO’s Index Translationum, a large database of existing translations of texts from numerous domains (Bosser 2000). The data for this study was limited to
literary
translations into
Polish
, a total of almost 18,000 individual editions of novels or collections of short stories by 8290 authors and 6582 translators from 155 languages. Editions of the same translations by the same translators were treated as separate entities, since this allows to discover authors, texts, translators and translations of particular cultural significance (or merely popularity). The network, prepared in Gephi (Bastian et al. 2009), used the number of editions of texts of an author translated by a translator as weights.

3. Simple measures

It is not surprising that a lion’s share of translations into Polish as recorded in the Index Translationum was from English-language literature: a total of 12190 editions, or 67.73%; this domination occurs in most cultures nowadays; in Poland, English has been the most-translated language since the 1930s (Krajewska 1972, 14). This is also visible in the most-translated (or, more precisely, most-published) authors: of the top 10, only Hedwig Courts-Mahler and Jules Verne wrote in a language other than English (respectively German and French); of the top 5, only Jack Higgins was male, preceded by Barbara Cartland (the top scorer with 75 published translations), Nora Roberts, Agatha Christie and Danielle Steel; the top 10 also included Lucy Maud Montgomery, Graham Masterton, André Norton and various Disney products (treated jointly here). Authors deemed as “classical”, “canonical” or “artistic” only began to appear at the bottom of the second ten: Philip K. Dick came as 19
th
, followed by Mark Twain, Jack London and Hans Christian Andersen; Dickens was 27
th
; Tolkien, 41
st
; Conrad, 46
th
; Ursula K. Le Guin, 54
th
, beat Shakespeare, 56
th
. German and French literature came far behind English and close to each other, with, respectively, 6.91% and 6.77% of the entire database, followed by Russian (4.22%), Spanish (2.16%) and Italian (1.90%). Apparently, Poland’s most prolific and ubiquitous translator is one Małgorzata Fabianowska, the translator of 73 editions by 71 authors (Disney products and books based on films), followed by Jacek Manicki, with 66 translators of such authors as Stephen King, Robert Ludlum and John Grisham. The fact than none of Poland’s celebrity translators appeared in any top positions shows the extent to which quantity does not equal quality in literary translation.

3. Social network analysis

Figure 1 presents a three-dimensional network visualization of the data, produced with the Fruchterman-Reingold force-directed algorithm (1991) and edited with UCSF Chimera (Pettersen et al. 2004).

Figure 1. Writer-translator network. Nodes are writers and translators; edges are translations.

As is evident from the general description of the content of the database, the large community in the middle of the sphere shows a strongly-interconnected web of translations from the English. By comparison, other major literatures (German, French, Russian, Spanish and Italian) are much less represented. There is precious little connection between the source languages (despite the above-mentioned existence of multilingual translators). The debris at the external limits of the sphere are instances of one-time translators of texts by authors translated only once, and thus unconnected with the community of a given source language; instances of a single author translated just twice or thrice and/or by just one or two translators, etc. Figure 2 presents an example of this:

Figure 2. A small community of two French authors, Peyremaure and Reznikoff sharing a translator, Olędzka; Reznikoff was also translated by Wasitowa.

Whenever larger communities are less interconnected, it is feasible to follow the connections within. Figure 3 does this for a subset of the French community. Jean-Paul Sartre is thus connected via his translators to such writers as Raymond Queneau and Jacques Prévert (both
Satrapes
of the Pataphysical Club), indicating that sharing translators might follow a chronological if not a generational or an ideological logic. But a somewhat more versatile translator, Jacek Trznadel, also links Sartre with Marquis de Sade. Numerous other interesting connections could be observed.

4
. Conclusions

Social network analysis thus seems a good tool to examine relationships within a large database such as the Index Translationum, relationships which are not accessible through direct retrieval of its information. The examples of connections mentioned above obviously do not even begin to describe the extent to which visualizing these networks can be helpful to write, perhaps, an entirely new history of literary translation. Once similar network are drawn for other target languages (and perhaps for more reliable databases of translations than the Index Translationum, which, for all its size, has some issues of reliability and representativeness), mathematical social network values (such as homophily, density, or clustering coefficient) could be compared between the literary cultures of these languages to produce quite a new type of distant reading of the phenomenon of literary translation.

Bibliography

Bastian, Mathieu, Sebastien Heymann and Mathieu Jacomy. 2009. “Gephi: an open source software for exploring and manipulating networks.”
Proceedings of the International AAAI Conference on Weblogs and
Social
Media
, San Jose, Ca.

Bosser, Sylvie. 2000. “The Index translationum on CD-ROM: an analytical tool.” In
World culture report 2000: cultural diversity, conflict and pluralism
, 274-275.

Fruchterman, Thomas M. J. and Edward M. Reingold, Edward M. 1991. “Graph Drawing by Force-Directed Placement.”
Software – Practice & Experience
21 (11): 1129-1164.

Hanneman, Robert A. and Mark Riddle. 2011. “Concepts and Measures for Basic Network Analysis.” 
The Sage Handbook of Social Network Analysis
. SAGE, 346–347.

Krajewska, Wanda. 1972.
Recepcja literatury angielskiej w Polsce w okresie modernizmu (1887-1918) : informacje, sądy, przekłady
. Wrocław: Ossolineum.

Pettersen, Eric F., Thomas D. Goddard, Conrad C. Huang, Gregory S. Couch, Daniel M. Greenblatt, Elaine C. Meng and Thomas E. Ferrin. 2004. “UCSF Chimera – a visualization system for exploratory research and analysis.” Journal of Computational Chemistry 13: 1605-1612.