The way people use basic words predicts how important they are going to become in online communities, indicates ongoing research.
KNOWLEDGE @ BI: Social Media Influence
Analyses of user-generated texts show that members who occupy different positions in online communities use pronouns very differently. These findings can be used to identify influential and mediators in online communities from the text people have written.
Online communities like Facebook, Twitter, Amazon Customer Discussions, EBay Discussion Boards and many others are heavily used by marketers for advertising, targeting, measurement of customer satisfaction and customer loyalty, complaint management and public relations.
- Read also: Even more hooked on social media
In search of influencers and mediators
Naturally, some of the members in these communities are more popular than others, and some are more likely to become mediators between other members. Influential and mediators are very important for processes like diffusion of information and word-of-mouth and identifying them early can make a big difference for the success of marketing efforts.
In some online communities users can follow or connect to other users and form social network (Facebook and Twitter, for example). Through analyzing the social network of the community, marketers can compute to what extent each member is influential and mediator.
However, it is often very difficult to observe all of the connections between members in a community. In addition, in most online communities members can not explicitly state which of the other members they are following or are friends with. In such communities, estimating how influential or mediating members are is difficult or relies on ratings that are very different from social influence.
Differences in writing style may be revealing
On the other hand, one of the most wide-spread and accessible online user-generated data is written text. Differences in writing style of users may be related to their positions in online communities. The relationships between use of different word categories and social network metrics are the topic of my doctoral project at BI Norwegian Business School, supervised by Professor Luk Warlop and Associate Professor Rutger Daniel van Oest.
The research project uses publicly available user-generated data from two large and very different online communities: epinions.com (product reviewing community with 51000 members) and gather.com (blogging community with 56000 members).
These communities give their members the opportunity to form public interpersonal connections. On one hand, the social network of each community was used to calculate influence and mediation metrics for the users as of present day.
On the other hand, all product reviews and blog posts that members have written in the first 90 days of their activity have been used to calculate to what extent people use different basic word categories.
Differences in the use of “I”, “we”, and “they”
The researchers concentrate on basic word categories like pronouns, negations, articles, prepositions, conjunctions, positive and negative words, tentative and certainty words and others because these are the most frequently used words and can not be avoided in speech and written text.
Analysis of the relationship between the use of some basic word categories in the first 90 days of members’ activity, and the influence and mediation metrics of the members at least 360 days since they became active, suggests that there are marked differences between the language people with different influence in online communities use.
Among all the differences, two findings about pronouns are most important. First, people with higher influence and mediation score use “we” words compared to “I” words significantly less often. Second, people with higher influence and mediation score use “we” words compared to “they” words significantly more.
Implications for marketers
The findings suggest that people who occupy different positions in online communities, use language differently and this difference can be detected on very low level, by simply counting basic words. Differences in using pronouns, which are the most frequently used word category, are strongly associated with different influence and mediation scores of members of online communities.
This implies that publicly available data like the number of certain words that people use in written text, can be used to estimate the extent to which they are likely to be or become influential and mediators in online communities.
On latter stage, marketers can use these estimates to improve their segmentation tools and to amend their campaign management by targeting members who occupy more important positions.
This article is published in BI Marketing Magazine Nr. 1-2014 (Link to E-Magazine). BI Marketing Magazine is a Science Communication Magazine published by the Department of Marketing at the BI Norwegian Business School.
Send your comments and questions regarding this article by E-mail to firstname.lastname@example.org