order of 1.6 or larger
87
The effect of any multiplicities of example sentences is a substantial decrease in the probability of zero
matches (dissimilarity) within a chosen headword, and a marked increase in the prospect of Kiddor Prinsloo
similarities. This increase will occur but be due only to the multiplicities; even if random matches are plausible
as in the circumstances outlined . . . [above]".
An example of what Prof Dunne is referring to is to be found in relation to the headword "spring" referred to
in paragraph 48 above.
[56] Reverting to the example sentences in respect of "baie", in the Pharos work the word has been referred to as
a noun (with 5 senses and 10 example sentences), an adverb (with 5 senses and 5 example sentences) and
a numeral with 1 example sentence. The OUP work distinguishes only an adverb and a numeral. The OUP
adverbs define 4 senses with 8 example sentences, and the numeral, 3 senses with 5 example sentences.
There are
Page 491 of [2015] 3 All SA 478 (WCC)
therefore 16 example sentences under "baie" in the Pharos work, and 13 in the OUP work.
[57] In the founding papers Dr Anton Prinsloo refers to 4 such example sentences in aspect of the headword
"baie":
57.1
Firstly, "I like Theo very much: he's a nice chap" in the Pharos works is compared with OUP's quote "I like him
very much". This is said by Dr Prinsloo to be an example of a direct copy, whereas it appears to be a sentence
wherein the subject is changed but the gist of the sentence remains the same.
57.2
Secondly, it is correctly pointed out that "thank you very much for your help!" in the Pharos work is taken up
identically in the OUP work.
57.3
Then "daar is nie baie melk oor nie" in the Pharos work is said to be identical to "daar is nie baie sap oor nie" in
the OUP work; and
57.4
Finally, Pharos' "sy het op haar verjaarsdag baie presente gekry", is said to be identical to the OUP's
"Nkosinathi het baie presente gekry".
[58] Assuming for the purposes of argument that all 4 OUP example sentences are indeed identical (which quite
arguably they are not), the fact remains that only four of Pharos' 16 example sentences have been copied by
OUP. This appears to equate to 25% of the Pharos work in respect of the headwork"baie".
[59] Prof Dunne offers the following initial comments in this regard.
"19.
It is feasible that many similarities emerge, because it is an intention of dictionaries to reflect common usage.
Some of these similarities might be more striking than others. Such extreme coincidences may admit several
possible single or multiple causative explanations. These explanations may involve prior occurrences, including
but not limited to frequency of usage, context of usage, common source corpora, commonly used expressions,
or indeed copying.
20.
Identical entries appeared to constitute a minimal subset within the reported similarities. This limited extent
obviates against inferences of simple lifting of text from one source to another.
21.
Similarities, as subjectively perceived will admit the same prior single or multiple causative explanations as are
applicable to extreme coincidence, such as in happenstance identical entries.
22.
Any vagueness of similarity criteria will involve or permit intrusion of retrospective elements driving towards
subjective declarations of near equavalence. Where the retrospective criteria are unspecified or implicit rather
than explicit and specific, they may be difficult to detect, communicate, replicate and verify.
23.
The criteria are the collective set of features whose presence within both dictionaries' entries for the same word
may lead an interested party to report either a similarity or dissimilarity.
24.
Different parties may adopt distinct criteria and their contrasting views of the same objective data source will
lead to distinct subjective overviews and inferences. To explore data appropriately it is not sufficient to
examine number and count alone, but also the criteria by which element of the count was derived, and the
connection of that element with the reality of its putative source."
Page 492 of [2015] 3 All SA 478 (WCC)
[60] And in conclusion, Prof Dunne offers his evaluation (from a purely statistical perspective) of the likelihood of
plagiarism by OUP or not:
"90.
Further similar remarks apply to EnglishAfrikaans as to AfrikaansEnglish structures. Firstly, both dictionaries
have substantially sized bodies of unique headwords. The limited common internal structures and the various
identified schooloriented purposes may give rise to natural similarities in the common segments. This natural
emergence of similarities will be exacerbated and appear more frequent than it really is if the analysis ignores
the multiplicities of senses for a single headword that necessarily reduce the probability of nomatch outcomes.
91.
At face value the two B D I S analyses suggest strong parallels of the two language sides, with respect to
unique and common parts. If plagiarism is to be sustained as an inference, there will also need to be a
plausible single explanation for plagiarism across both these B D I S sides in their entirety. The explanation will
have to deal with the current inherent twofold exaggeration of percentages for headword similarities arising
from multiple entries, and with the elimination of unique headwords from discussion. The explanation will have
to robustly defend the criteria of a very weak form of observable similarity (i.e. Prinsloosimilarity) as an
unassailable artefact of plagiarism alone, and the exaggerative effects of such loose criteria on the associated
counts and percentages reported.
92.
Such an explanation will need to address issues of motive, generality of alleged artefacts across compilers,