Hidden sentiment behind letter repetition in online reviews

Pak, Irina * and Teh, Phoey Lee * and Cheah, Yu N (2018) Hidden sentiment behind letter repetition in online reviews. Journal of Telecommunication, Electronic and Computer Engineering, 10 (3-2). pp. 115-120. ISSN 2180 1843

Text (Publisher's version is available) Teh Phoey Lee Hidden Sentiment Behind Letter Repetition.pdf - Published Version Restricted to Repository staff only Available under License Creative Commons Attribution Non-commercial. Download (1MB)

Abstract

Minimal research has been done on how letter repetition affects readers’ perception of expressed sentiment within a text. To the best of the researchers’ knowledge, no studies have tested samples of text with letter repetition using sentiment tools. The main aim of this paper is to investigate whether letter repetition in product reviews are perceived to have any sentiment value, based on ratings by individual participants and analyses using sentiment tools. This study collected and analysed 1,041 consumer reviews in the form of online comments using the UCREL Wmatrix system, and simulated emotional words within the comments to contain repeated letters. A group of 500 participants rated 15 positive comments and 15 negative comments and their respective simulated counterparts, while 32 sentiment tools are used to analyse a pair of positive comment and its simulated counterpart and a pair of negative comment and its simulated counterpart. Results indicate that readers perceive letter repetition to amplify a comment’s sentiment value, in which the effect was found more strongly in negative comments than positive comments. On the other hand, analyses using sentiment tools show that a majority of these tools are unable to detect letter repetition within a word and instead, treats the word as a spelling mistake. As consumers or online users, in general, have been found to use letter repetition to intensify and express their sentiments in their comments, this study’s findings suggest that letter repetition processing in any text-based mechanism needs to be enhanced. The outcome of this paper is useful for improving the measurement of sentiment analysis for the use of marketing applications.

Actions (login required)

View Item