ABSTRACT
As an emerging audience engagement channel for news organizations, news chatbots can interact with and attract audiences in a conversational manner. The present study applies the comparative digital journalism frameworks and examines how society-level factors-such as media systems and information communication technology's development-explain chatbot implementation on social media platforms. We surveyed 365 news organizations across 38 countries or regions and inspected their Facebook Messenger accounts with a mixed-methods approach. We found that less than half of the surveyed news organizations implemented Messenger, and only 67 Messengers were responsive-i.e. able to produce at least one response. We used the walkthrough method to interact with the Messengers with 22 pre-defined search queries on information seeking and navigation related to COVID-19. Then we used qualitative content analysis to examine the contents generated by the Messengers. Some Messengers are out of service or could only provide limited services (e.g. generating templated responses or closed-ended options). The Messengers in different news organizations demonstrated great variations in their capacity to understand the queries and interact with the audiences and reparative strategies to handle search failure. We proposed a three-category typology of news chatbots and offered practical and constructive suggestions for news organizations.
ABSTRACT
Social media has become a channel through which journalists distribute their work, reach audiences and gain visibility. Informed by the frameworks of journalistic branding, the heuristic-systematic model, and hypertextual elements, the present study examines the extent to which the source factor (journalists? branding on social media profiles) and message factors (communication styles and hypertextual elements) influence visibility (i.e. the popularity of the account and the number of favourites and retweets of the posts). We analysed the Twitter profiles of 98 health journalists from seven major media organizations in the US and conducted a manual content analysis of a representative sample of their public tweets (n?=?3982) published during the Covid-19 pandemic. In contrast to expectations, branding contributed little to any indicators of visibility, and profiles with institutional branding had fewer followers. Both affective messages and rational messages received more likes and retweets than messages without these elements. Tweets containing images or news-related hyperlinks received more retweets, whereas the number of @mentions in a tweet was negatively related to visibility. Journalists from traditional media, those who tweeted more often, and those with more followers had higher levels of visibility.