Abstract
We investigate two novel predictors of review helpfulness: grammatical classes of words, such as verbs and articles, and the arrays they form, or syntax. Patterns of grammar and syntax are known to co-occur with speech varieties that affect communication outcomes. However, measuring text syntax can be methodologically challenging. To address this, we analyze ordered word positions within reviews rather than full texts and look for text segments where the diversity of unique words and classes of words peaks. We find that the five-word segment at the beginning of a review, which we call the opening, exhibits these characteristics. Through statistical modeling and content analyses, we show that grammar and syntax classes in the opening predict review helpfulness and co-occur with specific clusters of words in the full text. Additionally, experimental studies provide evidence that the opening does not work in isolation, supporting the assumption that consumers read reviews in full. These findings may help to simplify online review analyses and inform future research agendas on consumer reviews.