SMART AWARD 2014 winner

LANGUAGE USE IN ONLINE REVIEWS.
A construal level perspective.

by Salina Smellers

In this master thesis we focused on the way in which abstraction is shown in online reviews and on the elements associated with this level of abstraction.

Firstly, we determined how abstraction can be measured in online reviews. For this, three parameters were identified: the abstraction score of the Linguistic Category Model (Semin & Fiedler , 1989), the number of words and the number of pronouns . These degrees of abstraction were used to determine which elements of reviews and review websites could have an impact on the level of abstraction. Based on our literature 12 hypotheses were drawn . These were tested using a content analysis of eight review websites. In total, 400 reviews were analyzed .

The results showed that both abstraction score, the number of words and the number of pronouns can be used as parameters of abstraction in online reviews. As reviews are more abstract (higher abstraction score), they contain fewer words and fewer pronouns. As to the elements that have an impact on the level of abstraction, we found for the reviewer that the presence of a user name is linked to more concrete reviews. Reviewers with some kind of status write more abstract reviews.

Also the subject of reviews affects the way reviewers describe their experience. Reviews on experience products were more concrete than reviews on search products. The presence of a product picture was associated with more concrete language. Regarding the influence of websites, our results showed that websites with product pictures on their homepage contain more concrete reviews. For the type of websites, there are indications that social websites offer more concrete reviews than non social websites .

Finally, instructions regarding the minimum length of reviews is closely linked to the level of abstraction of reviews. If websites require a minimum length, reviews will be more concrete. The procedure of writing a review is also associated with the level of abstraction in reviews. If reviewers must assign a rating first, reviews will be more abstract.