Is Language Diversity in Corporate Tweets the Online Rosetta Stone?

Synopsis The research problem This paper investigated the impact of language diversity in corporate tweets on the information asymmetry of firms to ascertain whether communication in different languages would cause noise and increase information asymmetry, or inform and reduce information asymmetry. Motivation Akin to the Rosetta Stone of ancient Egypt, current-day corporate tweets are written in more than one language to communicate with audiences of different languages. This leads to the question of whether the use of multiple languages assists in reducing information asymmetry between the company and its investors. The test hypothesis We hypothesized that a negative association exists between information asymmetry, proxied by cost of equity and bid-ask spread, and language diversity, proxied by a measure that accounts for both the number of languages used and the lexical distance of the most commonly used language in the tweets and the other languages used. We also examined the influence of emojis in corporate tweets. Target population We identified a data collection set of 1,420 major companies, consisting of 33 indices of 32 global exchanges and covering 36 countries from the year 2018. Of those, 758 companies from 34 countries have tweets. The language and emoji data from 1,671,030 corporate tweets in 95 different languages comprised the base data for our analysis. Adopted methodology The proxy for language diversity is based on the lexical distances between languages as proposed by Beaufils & Tomin ( 2020 ). A higher distance between the most used language and the other languages used for tweets is widely seen as a company’s attempt to reach an audience in a language group that is not familiar with the company’s primary language. If this approach lowers information asymmetry, then it would improve communication effectiveness. The sum of lexical distances for those languages constituted the language diversity score for each company. We also analyzed the percentage of tweets with emojis. Analyses Data were collected from tweets, corporate websites, and websites of relevant institutions using Python programs and processed using virtual machines rented from Amazon Elastic Compute Cloud. Ordinary least squares were used for the multivariate tests between the information asymmetry proxies (cost of equity and bid-ask spread) and language diversity and emojis. We controlled for other Twitter, firm, industry, and country variables that can affect information asymmetry. We also ran tests to account for the endogeneity of language diversity in the research models. Findings The costs of equity and bid-ask spreads of firms are negatively associated with language diversity and emoji use. The results of tests accounting for endogeneity and additional tests to gain further insights into the use of other languages supported the main test results Implications The study informs managers, policymakers, and researchers about the benefits of language diversity in corporate communication in a global setting. It also contributes to the nascent literature on multiple language use for external reporting and communication in accounting and business.