📄 To a dubious critical salvation: Etienne Leroux and the canons of South African English criticism

Summary

This article studies how Etienne Leroux, a leading Afrikaans novelist of the Sestigers, was received in English after translation. It centres on his novel Sewe dae by die Silbersteins (1962, Hertzog Prize 1964) and on Charles Eglington's English version, Seven Days at the Silbersteins (1964). As their framework, the authors adopt André Lefevere's theory of "rewriting", set out in his 1992 book on translation, rewriting and how literary reputations are manipulated. In that theory, translations, reviews and literary histories are all re-descriptions shaped by ideology, poetics and patronage. They first trace Leroux's turbulent reception at home, where church, state and volk recast his books as blasphemous or pornographic. They then follow how the translation passed his work into a different English South African patronage system. Their main claim is that English critics often read Leroux only in translation, cut off from the Afrikaans context. As a result, shaky judgements hardened into routinely repeated "wisdom", and the English assessment of him became cursory and patronising. The case matters as a warning that reputations built on translated texts can drift far from the original.

Key points

• The core case is Sewe dae by die Silbersteins (1962, Hertzog Prize 1964) and its translation by Charles Eglington. The CNA commissioned that translation in 1964 at the initiative of Philip Stein, who also had Eglington translate the sequel Een vir Azazel.
• The framework is Lefevere's "double control factor". The first control comes from professionals inside the literary system, such as critics, reviewers, teachers and translators. The second is patronage: publishers, censors, media and educational institutions. The authors add James F. English's economy of prestige, Bourdieu and Even-Zohar's polysystem theory.
• Charles Malan (1987) compared Leroux's domestic reception to the publishing history of Byron's Don Juan. In both cases the work was recast into "strange new texts" through press articles, letters and censorship findings.
• Early publishers rejected Leroux's work on moral grounds. For Die eerste lewe van Colet (1955), he had to personally guarantee sales of 1000 copies within three months, at a cost of 260 pounds.
• Leroux's English translations are now out of print. English criticism of his work focuses mostly on the political controversy around Sewe dae and Magersfontein, O Magersfontein! rather than on literary merit.

Interesting ideas

• I1. Each re-description of a text (a translation, review or summary) is treated as a new derived text carrying its own biases, not as a neutral copy of the source.
• I2. Judgements based only on a translation, read apart from the original-language context, can solidify into errors that later sources copy as settled fact.
• I3. Lefevere separates two filters on literature: internal experts and external patrons, which act on different grounds (poetics versus ideology).
• I4. Malan's commentary is described as rewriting the rewriters, meaning every interpretation layer imposes its own framework on the one before.
• I5. Repeated paraphrase across many secondary sources can create a false sense of consensus.

For Mova

How these ideas could serve Mova, our small open language model — its files, stages and tests:

• I1 (back): The text → graph → text compressor could label every regenerated text as a derived version and record which source graph it came from, so drift from the original can be measured.
• I2 (learn): When building silver data, witness parsers or LLM teachers that share one upstream source should not count as independent agreement, and negative controls should test for such inherited errors.
• I3 (world): The knowledge graph could store, for each evaluative claim, who asserted it and from which position, so the world of a text distinguishes facts from sourced opinions.
• I4 (frames): Pragmatic analysis of reviews or reported speech could nest attributions (A says B claims C), keeping each speaker's stance explicit instead of flattening everything into a single assertion.
• I5 (answer): QA readers could trace a frequently repeated claim back to its earliest source before treating repetition as evidence, and explain that trace in their answer.