Media 24 Books (Pty) Ltd v Oxford University Press Southern Africa (Pty) Ltd

Media24 Books published the Pharos Aanleerderswoordeboek vir Skole (Aanleerderswoordeboek), a bilingual Afrikaans-English dictionary for learners aged 10-16, first published in 1993 and republished in 2006. Oxford University Press Southern Africa (OUP) published a competing work, the Oxford Afrikaans-Engels/English-Afrikaans Skool Woordeboek (Oxford Woordeboek) in 2007, directed at the same market. In 2011, while preparing a new edition, Media24 examined the Oxford Woordeboek and concluded that it had been substantially copied from the Aanleerderswoordeboek. Media24's evidence focused on correspondences in example sentences in both dictionaries. Dr Anton Prinsloo examined four letters (B, D, I, S) on both sides of the dictionaries and identified correspondences ranging from identical sentences, to sentences with minor changes, to thematic similarities. Statistical analysis by Professor Kidd indicated 16-24% overlap in example sentences. OUP's defense included affidavits from three independent compilers (Mrs Aletta Cloete, Dr Liezl Potgieter, and Mrs Daphne Paizee) who denied copying and explained their methodology. Expert evidence from Professor Taljard and Mr Rundell explained that such correspondences were likely inevitable in basic bilingual dictionaries aimed at school children, given the limited vocabulary, simple sentence structures, and need to relate to children's life experiences. Professor Dunne's statistical analysis showed the correspondences were evenly distributed among all three compilers. Media24 proceeded by way of application rather than trial.

• Whether the Oxford Woordeboek infringed copyright in the Aanleerderswoordeboek by copying example sentences
• What degree of similarity between works is required to establish copyright infringement
• Whether the onus of proving copying was discharged on the basis of correspondences between example sentences alone
• Whether, in the absence of oral evidence and cross-examination, denials of copying could be rejected based solely on similarity between works
• The application of the Plascon-Evans rule in copyright infringement proceedings conducted by way of application

The appeal was dismissed with costs, including costs consequent upon the employment of two counsel.

In copyright infringement proceedings based on alleged copying of a literary work, the claimant bears the onus of proving both objective similarity between the works and a causal connection (actual copying) on a balance of probabilities. Where a prima facie case of copying is raised through proof of similarity and access, it may be rebutted by a plausible explanation of how the alleged infringing work was produced without copying. In application proceedings without oral evidence, the Plascon-Evans rule applies: the respondent's version must be accepted unless it constitutes bald or uncreditworthy denials or is palpably implausible, far-fetched or clearly untenable. In the case of reference works such as dictionaries, particularly basic works directed at specific audiences, natural constraints on content and expression may result in similarities without copying. Such similarities do not automatically establish copyright infringement. The court must examine all the evidence, not only similarities but also differences and evidence of independent creation, to avoid being misled by "similarity by excision." Where correspondences between works amount to only a small percentage of the whole, are randomly distributed, and are explicable by reference to the nature and purpose of the work, and where those responsible for the allegedly infringing work provide credible explanations of their independent methodology supported by expert evidence, the onus of proving copying is not discharged by the correspondences alone.

The court made several non-binding observations: (1) Copyright traps (deliberate errors inserted to detect copying) were noted as a method sometimes used by publishers of directories and dictionaries, but Media24 did not rely on this method. (2) The court noted that establishing plagiarism in a reference work is more difficult than in creative works like novels or songs because reference works are assemblages of generally available knowledge where correspondence between different works is expected. (3) The court observed that while unconscious copying is theoretically possible in copyright law, it is unlikely to explain copying of numerous example sentences in a dictionary compilation project, as lexicographers do not typically retain dozens or hundreds of such sentences in subconscious memory. (4) The court suggested that proper statistical evidence comparing the actual rate of correspondence with the expected rate in dictionaries of this type might have been helpful but was not provided. (5) The court commented that the allegations of copying during the editorial phase rather than the compilation phase were highly improbable and were not seriously pursued. (6) The court observed that it will only be in the most egregious cases that a court will reject a denial of copying merely on the basis of comparing two works without cross-examination of the author(s) of the alleged infringing work. (7) The court noted that OUP's attorneys had warned before proceedings commenced that the dispute was not appropriate for resolution by application and should proceed by way of trial, but Media24 proceeded by application nonetheless.

This case is significant for South African copyright law in several respects: (1) It clarifies the approach to establishing copyright infringement in reference works such as dictionaries, which involve assemblages of generally available knowledge where some similarity is expected. (2) It emphasizes that while similarity and access may raise a prima facie case of copying, the copyright holder bears the onus throughout and must establish copying on a balance of probabilities. (3) It highlights the limitations of pursuing copyright infringement claims by way of application without oral evidence, particularly where denials of copying are supported by plausible explanations and expert evidence. (4) It demonstrates the application of the Plascon-Evans rule in copyright cases, requiring acceptance of respondent's version unless it constitutes bald denials or is palpably implausible. (5) It warns against the danger of "similarity by excision" - focusing only on similarities while ignoring differences and original work. (6) It establishes that in works directed at specific audiences (such as school dictionaries), constraints on content and expression may naturally lead to similarities without copying. (7) It indicates that claims of unconscious copying are unlikely to succeed where the extent of alleged copying is substantial and would require retention of numerous sentences in memory.