Samineh Razeghi; Farhad Sasani; Marjan Taheri
Abstract
Every part of a conversation is related to what comes before or after, however, there is sometimes a long gap between the beginning and the end topic of the conversation. Hobbs (1990) calls this phenomenon “topic drift” and defines 4 types of coherence relations responsible for topic drift. ...
Read More
Every part of a conversation is related to what comes before or after, however, there is sometimes a long gap between the beginning and the end topic of the conversation. Hobbs (1990) calls this phenomenon “topic drift” and defines 4 types of coherence relations responsible for topic drift. Since the messages are displayed in a reverse chronological order and subsequently it is not obvious to which message the comment refers, maintaining coherence in multi-user computer-mediated communication is rather difficult. Moreover, intervening messages are observed among the messages that logically form a chain. As a result, one can consider the existence of coherence relations and analyze topic drift in conversations occurring on social media. In this paper, we have attempted to use Hobbs (1990) and Herring and Nix’s (1997) framework to investigate the coherence relations among 27 posts and 4881 comments on Instagram. In comments related to the main post, the order of coherence relations was “on topic and semantic parallelism” with the distances 3, 2 and 1 respectively, “break” with the distance 4, “explanation” with the distances 3, 2, and 1, and “metatalk” with the distances 0, 4, 3, 1, and 2. In comments forming intervening chains, the order of coherence relations can be defined as: “on topic” with distance 0, “explanation” with distances 1, 3 and 2, “Semantic parallelism” with distance 1, 3 and 2 and “metatalk” with distance zero and 2.
