Analyzing chat experienced through the lens of Conversation Analysis - Part I

At Sendbird, we take research seriously to ensure that our product decisions are driven by data. Whether through primary or secondary research, we strive to develop a deep understanding of human conversational needs so that our products can better serve those needs.

As a part of this effort, Sendbird’s Research team has reviewed the literature in the field of Conversation Analysis (CA), which is recognized as one of the most effective ways to study human conversations. In this blog series, we will introduce the fundamental patterns of conversations from CA and discuss how those patterns are manifested in today’s chat experiences. We hope that you find value in this series.

Nature of chat

In online chat, both written and spoken language converge in a single medium. Written communication is asynchronous in nature, due to the delay between interactions, while spoken communication is almost entirely synchronous as it takes place in real time. Therefore, online chat is considered quasi-synchronous— a combination of asynchronous written language and synchronous spoken language.

Since chat is a form of written language, it has some benefits over other traditional methods that rely on oral conversations such as face-to-face interactions or phone calls. Typing into chat requires less emotional investment as it is less personal. It also leaves a trace of the conversation where both parties can retrieve it later, enabling responding at a later time.

The synchronous aspect of chat conversation mirrors how people talk at its core. Conversations are co-created between two speakers, coordinated in real time. In his book Using Language (1996), psycholinguist Herbert H. Clark described conversation as “a form of joint action” similar to the coordination that happens when “two people waltzing, paddling a canoe, playing a piano duet, or making love.” Therefore, to enhance your businesses’ live chat experience, it is essential to have a deeper understanding of the fundamentals of human conversation.

Conversational Analysis

Conversational Analysis (CA) is the study of fundamental patterns of interaction in human conversations. Emerged from sociology in 1970, CA has been used as a method to analyze both the spoken and written interactions in various fields. More recently, it has been applied to analyze quasi-synchronous interactions in online chat as well as to inform chatbot design.

Regardless of whether you are planning to implement or have already implemented chat in your app, CA can provide a useful conceptual framework that gives a deeper understanding of your users’ conversational needs. In this blog series, we will introduce you to key concepts from CA and relevant real-world best practices.

Turn-taking

CA views conversations as a series of turn-taking, embracing both verbal and nonverbal expression. In oral conversation, the turn-taking is so finely tuned that people start speaking within microseconds after their partner finishes. This gap between the lines is called transition relevance place (TRP), a place a listener looks for to take their turn.

Online chat conversations follow the same model of turn-taking. All messages are sent and received sequentially; however, it lacks traditional nonverbal cues. Therefore, recognizing your TRP is more challenging and tends to take longer.

As a way to overcome this limitation, messaging apps and chat platforms have been adopting features that assist users to recognize TRP. The typing indicator is one of the examples. This feature allows users to know visually if another user is typing a message. It remains visible until the user sends the message or deletes the text completely. That way, users can determine the right timing for them to start their turn.

The read receipt is another example of a feature that assists users to recognize TRP. A read receipt shows a sender when a message has been read by the other user. This feature allows the sender to decide whether it makes sense to maintain their turn in the conversation or not.

Until the message is marked ‘read,’ the sender is more likely to wait until it’s been read by the other user instead of sending more messages. On the other hand, if the message was left on read (marked ‘read’ but unanswered), the sender is more likely to maintain their turn to clarify the previous message or follow up with additional information.

Sendbird offers both typing indicators and read receipts for your in-app chat, in addition to numerous other features that are essential to optimize your chat experience. Learn why industry leaders like Reddit, the 19th most visited web app in the world, and Paytm, with over 330M total users, choose Sendbird’s conversations platform to power their in-app chat.

In the part 2 of this blog series, we will discuss different types of turn-taking cues. Stay tuned!