A fascinating investigation in the way scientific papers have been written over the past 4 decades has been carried out by Vinkers et al and the results were published in The BMJ at the end of last year.

The authors analysed the yearly usage of positive, negative and neutral words in PubMed abstracts from 1974–2014 and compared them with 4% of all books printed and digitised. The hypothesis was that the language used for scientific writing may have evolved to be more positive/negative as the push for innovation and throughput has increased.

25 words from each category were quantified and all 25 positive words contributed to an astonishing relative increase of 880% over the time period, although some words, such as ‘robust’, ‘novel’, ‘innovative’ and ‘unprecedented’, were more frequent than others. The increase in negative words was not quite so pronounced at 257%. This compared with no observable increase of neutral words or positive words in published books.

The authors concluded that the most likely explanation for this shift in language is that exaggeration and positive bias are perceived to be necessary to get results published. This is a worrying trend and the authors suggest that a reform of academic culture from quantity of publications to quality is needed.