The full document can be found on the following website:
http://social.cs.uiuc.edu/papers/pdfs/interact2005-kalra.pdf
TextTone: Expressing Emotion Through Text
Abstract. An increasingly large part of online communication is inherently social in nature. This social interaction is limited by the modalities of online communication, which do not convey tone or emotion well. Although some solutions have evolved or have been proposed, they are inherently ambiguous. We present a system, TextTone, for the explicit expression of emotion in online textual communication. TextTone incorporates reader-specific preferences for the visualization of tone and emotion. We describe social interaction and visualization scenarios that TextTone can be meaningfully used for and discuss an initial implementation. Finally, we present the results of a preliminary evaluation of this implementation of TextTone.
1 Introduction
An increasingly large part of online communication is inherently social [11]. Blogs, instant messaging, IRC chat rooms and the like are new forms of social communication, and even predominantly non-social interactions online have a significant social component [6, 8].
However, the modalities of online social interaction are much more limited than those of face-to-face social interaction, and provide no good way to convey non- verbal cues [3], which make up 93% of face-to-face interaction [7]. Online communities have appropriated the use of emoticons and other textual representations to help address this issue, but the lack of a standard set of expressive and versatile representations introduces ambiguity, which limits their usefulness [10].
TextTone aims to provide a means for unambiguously expressing emotion through text. It does this by dynamically switching representations based off of the preferences of the reader. Instead of forcing all the users to ‘speak a common language’, TextTone picks the most meaningful representation based on the audience, the environment, and the platform, so that two users reading the same text would each see the representations that made the most sense to them in their own environment.
2 Current Representations
The inability to convey emotion through text greatly limits the effectiveness of textual communication from a social perspective. Not surprisingly, therefore, numerous attempts have been made to address the issue, both by interaction researchers and the online community [1, 4, 5, 9, 13]. However, expressive representations, such as
M.F. Costabile and F. Paternò (Eds.): INTERACT 2005, LNCS 3585, pp. 966 – 969, 2005. © IFIP International Federation for Information Processing 2005
Abstract. An increasingly large part of online communication is inherently social in nature. This social interaction is limited by the modalities of online communication, which do not convey tone or emotion well. Although some solutions have evolved or have been proposed, they are inherently ambiguous. We present a system, TextTone, for the explicit expression of emotion in online textual communication. TextTone incorporates reader-specific preferences for the visualization of tone and emotion. We describe social interaction and visualization scenarios that TextTone can be meaningfully used for and discuss an initial implementation. Finally, we present the results of a preliminary evaluation of this implementation of TextTone.
1 Introduction
An increasingly large part of online communication is inherently social [11]. Blogs, instant messaging, IRC chat rooms and the like are new forms of social communication, and even predominantly non-social interactions online have a significant social component [6, 8].
However, the modalities of online social interaction are much more limited than those of face-to-face social interaction, and provide no good way to convey non- verbal cues [3], which make up 93% of face-to-face interaction [7]. Online communities have appropriated the use of emoticons and other textual representations to help address this issue, but the lack of a standard set of expressive and versatile representations introduces ambiguity, which limits their usefulness [10].
TextTone aims to provide a means for unambiguously expressing emotion through text. It does this by dynamically switching representations based off of the preferences of the reader. Instead of forcing all the users to ‘speak a common language’, TextTone picks the most meaningful representation based on the audience, the environment, and the platform, so that two users reading the same text would each see the representations that made the most sense to them in their own environment.
2 Current Representations
The inability to convey emotion through text greatly limits the effectiveness of textual communication from a social perspective. Not surprisingly, therefore, numerous attempts have been made to address the issue, both by interaction researchers and the online community [1, 4, 5, 9, 13]. However, expressive representations, such as
M.F. Costabile and F. Paternò (Eds.): INTERACT 2005, LNCS 3585, pp. 966 – 969, 2005. © IFIP International Federation for Information Processing 2005
TextTone: Expressing Emotion Through Text 967
kinetic or animated text, have met with limited success in practice, largely due to difficulties in interaction and archival [1, 4, 5]. Our system overcomes these kinds of issues through an easy-to-use authoring environment (Figure 1) and representations that can be naturally archived.
Currently, the most successful representation by far is the use of emoticons [9, 13]. However, studies have shown that only four emoticons account for almost all of the common emoticon usage online, with the prototypical smiley face – :) or :-), accounting for more than half [9]. This suggests that the set of emoticons that can be widely used is very limited. Additionally, different cultures sometimes adopt different representations for the same emotion – the textual representation of ‘angry’ in Japan is >.< , and is very different from the :@ commonly used in the United States1. Consequently, there is an inherent ambiguity in the usage of all but a very few emoticons, and expressing any kind of non-trivial emotion textually is unfeasible with these representations.
3 TextTone
TextTone attempts to address these issues by dynamically picking a representation that the reader has chosen to be meaningful and appropriate for that emotion, in that environment and on that platform. The author of a body of text semantically indicates tones in the text – much as emoticons are used today, and TextTone represents this text in an intermediate plain-text format that can be distributed to readers. When viewed, TextTone transforms the text in this format into the appropriate representation chosen locally by each reader. For example, a blogger could indicate that a certain line was written in anger, and everyone who read that line would see it represented to them in a way that indicated anger to them personally – be it with bold text, or red text, or large text, or whatever representation meant ‘anger’ to them.
This allows the author to select the tones that need to be conveyed, but delays the encoding of those tones until the text is viewed by the reader, thereby allowing the system to use each reader’s preferred representation.
As an initial exploration into the space, TextTone was implemented as an IM client based on the open-source DAIM library [12] for the AOL Instant Messaging (AIM) architecture. We felt that an IM client would allow us to study a wide enough variety of textual social interactions to be able to meaningfully evaluate the system.
Our implementation allows the user to connect to the AIM network and to communicate with people on their buddy list. In their instant messages, users can indicate tone by demarcating blocks of text with the corresponding emoticon. Alternatively, users can select text and choose the tone from a drop-down menu. The preset tones are: happy, very happy, upset, disappointed, angry, very angry, shocked, confused, winky, tongue-in-cheek, embarrassed and ‘none’ (the default no-tone option). The corresponding emoticons are also displayed. These tones were chosen as they represent the emoticons most commonly included with commercial IM clients. In
1 In reality, :@ is not always used to represent angry. We found other emoticons for angry (and other meanings for :@) on different systems. This only serves to further illustrate our point.
kinetic or animated text, have met with limited success in practice, largely due to difficulties in interaction and archival [1, 4, 5]. Our system overcomes these kinds of issues through an easy-to-use authoring environment (Figure 1) and representations that can be naturally archived.
Currently, the most successful representation by far is the use of emoticons [9, 13]. However, studies have shown that only four emoticons account for almost all of the common emoticon usage online, with the prototypical smiley face – :) or :-), accounting for more than half [9]. This suggests that the set of emoticons that can be widely used is very limited. Additionally, different cultures sometimes adopt different representations for the same emotion – the textual representation of ‘angry’ in Japan is >.< , and is very different from the :@ commonly used in the United States1. Consequently, there is an inherent ambiguity in the usage of all but a very few emoticons, and expressing any kind of non-trivial emotion textually is unfeasible with these representations.
3 TextTone
TextTone attempts to address these issues by dynamically picking a representation that the reader has chosen to be meaningful and appropriate for that emotion, in that environment and on that platform. The author of a body of text semantically indicates tones in the text – much as emoticons are used today, and TextTone represents this text in an intermediate plain-text format that can be distributed to readers. When viewed, TextTone transforms the text in this format into the appropriate representation chosen locally by each reader. For example, a blogger could indicate that a certain line was written in anger, and everyone who read that line would see it represented to them in a way that indicated anger to them personally – be it with bold text, or red text, or large text, or whatever representation meant ‘anger’ to them.
This allows the author to select the tones that need to be conveyed, but delays the encoding of those tones until the text is viewed by the reader, thereby allowing the system to use each reader’s preferred representation.
As an initial exploration into the space, TextTone was implemented as an IM client based on the open-source DAIM library [12] for the AOL Instant Messaging (AIM) architecture. We felt that an IM client would allow us to study a wide enough variety of textual social interactions to be able to meaningfully evaluate the system.
Our implementation allows the user to connect to the AIM network and to communicate with people on their buddy list. In their instant messages, users can indicate tone by demarcating blocks of text with the corresponding emoticon. Alternatively, users can select text and choose the tone from a drop-down menu. The preset tones are: happy, very happy, upset, disappointed, angry, very angry, shocked, confused, winky, tongue-in-cheek, embarrassed and ‘none’ (the default no-tone option). The corresponding emoticons are also displayed. These tones were chosen as they represent the emoticons most commonly included with commercial IM clients. In
1 In reality, :@ is not always used to represent angry. We found other emoticons for angry (and other meanings for :@) on different systems. This only serves to further illustrate our point.
968 A. Kalra and K. Karahalios
Fig. 1. The interface used to assign Fig. 2. An archive of one side of a representations to tones in TextTone, conversation on TextTone. The word showing the ‘very happy’ tone ‘awesome’ uses the ‘very happy’ tone
choosing a representation for emotion, users can control the text’s font-size, font- color, and font-face, and whether the text is bold, italicised, underlined or has a strike-through (Figure 1).
Fig. 1. The interface used to assign Fig. 2. An archive of one side of a representations to tones in TextTone, conversation on TextTone. The word showing the ‘very happy’ tone ‘awesome’ uses the ‘very happy’ tone
choosing a representation for emotion, users can control the text’s font-size, font- color, and font-face, and whether the text is bold, italicised, underlined or has a strike-through (Figure 1).
No comments:
Post a Comment