Promotion in science has attracted increasing attention in recent years, and various linguistic devices such as boosters, hypes, or positive words were used to promote research for publication purposes. The present study examined how researchers promote their work with linguistic devices such as positive and negative words and is probably the first one that investigates such promotional devices in not only abstracts but also full texts based on a large dataset of articles (more than 2.2 million). The results showed that the use of positive words in both the abstracts and full texts exhibited a significant uptrend in the past two decades, and positive words were more frequently used than negative ones in both abstracts and full texts. However, journal impact was found to be only very weakly correlated with the use of positive and negative words in both abstracts and full texts. The preference of promotional linguistic devices such as positive words may be explained by factors such as the increasingly competitive selection process of publication, the publication bias, the outcome reporting bias, and the universal positive bias. The study argues for a fair use of such promotional linguistic devices.