Design for Conversation

  • 8,052 views
Uploaded on

The popularity of social media focuses our attention on designing for conversations. Increasingly we create social digital spaces that are instant, ambient, and ubiquitous. What are the conversational …

The popularity of social media focuses our attention on designing for conversations. Increasingly we create social digital spaces that are instant, ambient, and ubiquitous. What are the conversational resources that we use in offline conversation that we need to take into account when designing these spaces? What is the difference between an online and offline conversation, if any? These are some of the questions we will consider as I review what ethnomethodologists know about how people converse in person and in technologically mediated contexts.

More in: Design
  • Full Name Full Name Comment goes here.
    Are you sure you want to
    Your message goes here
    Be the first to comment
No Downloads

Views

Total Views
8,052
On Slideshare
0
From Embeds
0
Number of Embeds
2

Actions

Shares
Downloads
0
Comments
0
Likes
9

Embeds 0

No embeds

Report content

Flagged as inappropriate Flag as inappropriate
Flag as inappropriate

Select your reason for flagging this presentation as inappropriate.

Cancel
    No notes for slide

  • I'm going to talk about a particular qualitative research method, Conversation Analysis, its history, how it has been used for the design of systems in the past and how we can use it to design better user experiences now. To begin my discussion I'm going to briefly summarize its history within the social sciences for those of you unfamiliar with it, show you an example of how it’s done, then discuss how design researchers have used it to construct and assess technological systems.

    How many of you are familiar with conversation analysis?
    Have any of you used it for design?

  • Those of you familiar with design research methods such as contextual inquiry and ethnographically informed design know that there’s a distinction between such approaches and the methods that incorporate cognitive science. Most theories of cognition suggest enduring qualities that shape situations.
    Ethnographically informed design instead considers the situation and how it shapes interaction. Conversation analysis is yet another way that we may consider situated actions in designing the user experience.

    Conversation analysis is aligned with the program of theory and research known as ethnomethodology. Ethnomethodology examines the "nature and origins of social order. It rejects ‘top-down’ theories that attribute the organization of everyday life to cultural or social structural phenomena understood as standing outside of the flow of ordinary events. Instead it makes use of a 'bottom-up' approach, and seeks to “recover social organization as an emergent achievement that results from the concerted efforts of societal members acting within local situations" (Maynard & Clayman, 2003). Social order does not proceed from a set of stable facts; instead, social action is achieved every day by independent individuals acting in concert.

    It considers practical reasoning and everyday common sense as an object of study. Specifically, ethnomethodology articulates the “taken-for-granted” rules of order that people use when negotiating everyday matters; the sorts of things that seem obvious to the actual participants in the situation. Ethnomethodology began in the mid part of the twentieth century through the work of sociologists who then began the program of study known as Conversation Analysis.

  • Initially, naturally occurring conversation was considered by such linguistic scholars as Chomsky to be too messy to be studied in detail. Yet Harvey Sacks, as he listened to recordings of phone calls to a suicide hotline, noticed the social function certain rhetorical moves served and he began to develop an empirical method for investigating conversational sequences. Unlike other fields that studied communication, conversation analysis examines naturally occurring talk instead of an idealized form.

    Recordings are important because people vary words with pitch rises and falls, etc. The history of conversation analysis is tightly tied with technology. It wasn't until audio and video recording became common that researchers were able to inspect natural conversation closely.

    Despite the early attention to recordings of phone calls, Conversation Analysis includes both linguistic and paralinguistic features of conversation such as speech, gesture and gaze. It's the study of talk-in-interaction and asserts that talk is used for something, it is functional. Talk performs work. Talk makes things happen and by using conversation analysis, we begin to uncover how.


  • One thing I want to stress is that what conversation analysis uncovers is something as capable members of society, who are familiar with our local situations, we demonstrate to each other every day that we already know how talk works. We tacitly orient to these norms. Conversation analysis helps us to make them explicit as we articulate these patterns so we can consider how to better design social media spaces.

  • As you know from your work, it’s usually when something goes wrong that the mechanism for the right path becomes apparent. When things go smoothly, the rules seem invisible, as you can't imagine things happening any other way. So conversation analysts study such things as repairs of conversational misunderstandings – those moments in a conversation when you realize that your interactional partner assumes you meant one thing when you meant another. Or when half way through a conversation your realize you misunderstood a crucial statement she made at the beginning. The work you do to remedy this problem is known as a repair.

    Importantly this sort of analysis involves only what the participants themselves can see and hear without addressing inner feelings or motivations.
    The meaning of a conversation is not found in language rules and dictionary definitions or as something that exists within the individuals; instead, it is actively negotiated among participants.

    It is the person who takes the next turn at talk who determines what interactional meaning the prior statement or “utterance” has. It’s a fundamental CA principle that interactions are structured as sequences in which each move incorporates the participants interpretation of the immediately preceding move (Wasson, 2000).

  • [Sound File]

    Before we move on to how to apply conversation analysis to design I’m going to spend just a few minutes showing you what the typical output of CA research looks like to make sure we all a good understanding of it.

    Continuing with the idea of repairing misunderstandings, here is a brief exchange between two sisters: Agnes and Portia on the phone. Portia starts the close of the conversation but it is taken by Agnes as a complaint about her not having time to see her sister.
    Portia utters the repair-initiating component “Oh” which is followed by an acceptance of Agnes’s excuse for not visiting.
    Portia: Well, honey? I'll probably see you one of these daysAgnes: Oh god yeah, I just couldn’t get down there.Portia: Oh- Oh I know. I'm not asking you to come down.

    In analyzing this bit of talk we note how Agnes received Portia statement but even more importantly we can see how Agnes reacted to Portia’s reception of her original statement. We can see the unfolding of shared understanding or “intersubjectivity.”

    So for conversation analysts such a brief conversation is rich with data about how we perform work with talk.

  • Just like the highly descriptive very local ethnographies of particular situations that you might be more familiar with, conversation analysis involves the detailed study of situated local action. But unlike a lot of qualitative research, through observation, conversation analysts attempt “to find patterns that can be formulated as normative behaviors” (Laursen, 2005). They gather multiple instances of the same conversational work in similar situations. These instances are generalizable and that is one of the things that makes this situated research so useful for design.

  • So, when Christian and Erin, in their book Designing Social Interfaces: suggest we should “write like a human” and HCI researchers recommend that computers should "conduct a dialog in as human like way as possible," conversation analysis offers a body of research that explains what exactly that means.

  • Lucy Suchman was one of the researchers who in the mid 80s made use of conversation analysis by considering the computer an interactional partner in a conversation. They used conversational analysis to study the order of conversation between a user and computer (Suchman, 2007).

    Other HCI researchers have suggested that conversation analysis provides a rigorous method for analyzing usability sessions (Douglas, 1995; Nieminen, Karjalainen, Riihiaho, & Mannonen, 2009).

    What exactly goes on between the facilitator and the user who is supposed to articulate what they are thinking as they attempt a task? Some researchers suggest that CA gives us a way to analyze this conversation and note the trouble spots.
  • Zawinski's Law of Software Envelopment
    “Every program attempts to expand until it can read mail. It’s a snarky comment by programmer Jamie Zawinski about software bloat. But there’s a reason that mail is what these programs eventually incorporate. People don’t want to talk with computers they want to talk with other people.

  • The literature that addresses computer-mediated communication includes a substantial body of work that considers text-based, synchronous computer-mediate communication and language use. In particular, IRC and chat within text-based virtual worlds such as MUDs and MOOs in the 1990s received the attention of researchers. These researchers considered the communicative practices in such environments and noted that they exhibit characteristics of both oral and written communication. (Werry, 1996)

    Some of you may be familiar with Walter Ong and his assertion that we’re entering a secondary orality made possible through these new means of communication. However, online conversation differs from ordinary conversation in a number of ways. Ordinary conversation relies on the allocation of turns. Online conversation does not have overlaps or interruptions (Werry, 1996, p. 51). Instead, conversational participants construct each utterance in isolation. They send each of them to the system that then posts them in the order received.

  • Previously the mathematical model of communication dominated engineering. And even as recently as this past summer in the ACM magazine, Interactions, the authors suggested that designing for conversation could simply be modeled on Shannon’s model of communication. In this model there is the information source and the signal that must be transmitted through varying levels of noise to the receiver. Although highly useful for developing communications systems, it’s not so useful when designing for conversation. It leads to some seemly practical yet unsatisfactory design decisions.

    For example, you might think that if your users repeatedly ask other users for some bit of information, then it makes sense to add a feature that makes that information available. However, their asking might be doing other work than just eliciting information. Without thoughtful consideration of what interactional work that question is doing you risk robbing them of that conversational resource.

    One early enhancement to text-based chat was providing users a way to set up a public profile where they could enter their age, sex and location. However, researchers found that the most common question in the chat continued to be “what is your age, sex or location?” Or sometimes a variation that revealed they had viewed that person’s profile, but they asked are you really in “Florida?” or whatever the user’s location was. Those questions were conversation starters.

    In designing for conversation, it’s important to observe the interactional work occurring, not just the lexical meaning. Don’t treat conversation as a pure exchange of information.

  • There haven’t been many guides written on how to actually integrate conversation analysis into the design process. However, one PARC design team did bring in a conversation analyst to work with them on a project and they’ve published their experience.

    The project was a descriptive audio system for a museum and they were particularly interested in fostering a social experience for visitors.
    An initial prototype used a speaker system so the experience could be shared. Afterward, they learned from user interviews that visitors valued both the shared experience with their friends and being able to move at their own pace. So the design challenge was to satisfy two conflicting goals. The conversation analyst was able to analyze the interaction with the first prototype to inform the design of the second. Specifically she learned that visitors allocated the audio a turn making it a conversational partner with themselves. Understanding how visitors to the museum integrated the audio into their personal conversations with the person who accompanied them was key to developing the second more sophisticated prototype that actually allowed visitors to
    •independently listen to the audio narration
    •listen in on what their friend was currently listening to
    •and talk with their friend, even over the audio

    This next iteration was successful and met the conflicting design goals. It was the nuanced understanding of interaction that allowed them to make appropriate design decisions. They analyzed the organization of talk with the second prototype and noted that users fluidly went back and forth between independent and social experiences continuing to incorporate the audio as if it were an interactional partner in the conversation.

  • In order to establish that a conversation is taking place there must be “coherence.” Coherence is marked by sustained, topic-focused, person-to-person exchanges." These person-to-person exchanges require some way of indicating whom we are addressing. In ordinary conversation, there are multiple ways of doing this including just a simple gaze.

    In the paper, Beyond Microblogging, the authors discuss user appropriation of Twitter for conversation. They ask "[h]ow well does Twitter support user-to-user exchanges?" In particular they consider the function of the @ sign as a marker of addressivity. In their data they found that there was "short, dyadic exchanges relatively often, along with some longer conversations with multiple participants that are surprisingly coherent. These conversations are facilitated in large measure by use of the @ sign as a marker of addressivity (i.e., to direct a tweet to a specific user) and the ability to “follow” other users, which aid users in tracking conversations."

  • In ordinary conversation, two-person or dyadic conversation is fairly straightforward with speaker and recipient constantly swapping roles back and forth. Multi-party conversation though is particularly interesting where the participants work constantly to build the identities of who is speaker and who is recipient again and again. The participants constantly modify the participation framework.

    The speaker can designate the next speaker through direct address, or gaze or a number of other actions depending on local practice. So in multiparty conversation a lot of work must be performed to maintain coherence and continually build the participation framework.

    Shared attention.

    On Twitter we lack many of these conversational resources which is why the researchers described finding multiparty conversations on Twitter “surprisingly coherent.”

    In one example from a six person conversation over several turns, the participants used both topicality and addressivity.

  • Example 1

  • Example 2

  • Shifts in footing can also be a problem with Twitter.
    Goffman introduced the footing concept in order to explore the nature of involvement and participation in social interaction (Clayman, 1995). Speakers can use interactional resources to show that their words are not entirely their own. For example, a journalist might explicitly say to a politician “your political opponent suggests that you’re soft on crime.” However, journalists also conduct interviews where they articulate a question that suggests a notion not their own without attribution, such as “Mr. President, will the health care bill place an undue burden on small businesses?” In the course of an interview, a journalist may shift footing several times, animating his own words and then articulating those of others’ to display provocative viewpoints. Participants on twitter may echo or paraphrase others’ tweets by attributing them using RT or via and the original author’s username. However, when giving voice to statements outside of Twitter, shifts in footing lead to confusion.

  • An example from the data that makes up the posts to twitter concerning the contested Iranian election this past summer demonstrates this confusion.

    Just to sum up the event, people in Tehran were calling in to a Persian radio station to report a possible strike in the bazaar, a Twitter user was listening...translating the statements into English and posting them on Twitter. It was over a period of a few hours. Those who came late had missed any explicit attribution for the claims being made and questioned him, accusing him of being the author of these statements and not just the animator.

  • Bob Moore, a game designer at Multiverse, notes that in virtual worlds quasi-synchronous chat lags behind synchronous avatar motion. Moore explains that “… a key feature of real-life conversation is that you can hear a turn unfolding in real time. This enables you to do things like determine who should speak next, anticipate precisely when the turn will end so you can start your next turn with minimal gap and overlap, and even preempt the completion of the current speaker’s turn if you don’t like the direction in which it’s going. In other words, the ability to monitor other people’s turns in-progress is a requirement for tight coordination in conversation. Most virtual worlds use IRC- or IM-style chat systems and therefore do not allow players to achieve this tight coordination among their turns at chat and avatar actions. The result is an interactional experience that feels very unnatural (at first) and that motivates players to invent workarounds to the system.”
  • Much like ethnography, conversation analysis uncovers opportunities for design. It is a technique for organizing our observations to derive models.
    So we’ve discussed the history of CA and how it’s traditionally done. We’ve also briefly gone over how to use it in iterative design. I’ll just leave you with three things to think about as you observe conversations in the systems you design.
    •Notice user “workarounds.” We discuss this a lot, often calling it desire lanes or paving the cow paths.
    •Observe the interactional work occurring, not just the lexical meaning.
    •Don’t treat conversation as a pure exchange of information.




Transcript

  • 1. IA Summit 2010 Design for Conversation or some troubles with Twitter Tanya Rabourn @Rabourn tanya.rabourn@gmail.com flickr: camera_obscura
  • 2. IA Summit 2010 Tanya Rabourn Conversation Analysis • History • Example • How it’s used in design flickr: sveinnbirkir 2 I'm going to talk about a particular qualitative research method, Conversation Analysis, its history, how it has been used for the design of systems in the past and how we can use it to design better user experiences now. To begin my discussion I'm going to briefly summarize its history within the social sciences for those of you unfamiliar with it, show you an example of how it’s done, then discuss how design researchers have used it to construct and assess technological systems. How many of you are familiar with conversation analysis? Have any of you used it for design?
  • 3. IA Summit 2010 Tanya Rabourn Background & History • Conversation Analysis • Ethnomethodology flickr: nasrulekram 3 Those of you familiar with design research methods such as contextual inquiry and ethnographically informed design know that there’s a distinction between such approaches and the methods that incorporate cognitive science. Most theories of cognition suggest enduring qualities that shape situations. Ethnographically informed design instead considers the situation and how it shapes interaction. Conversation analysis is yet another way that we may consider situated actions in designing the user experience. Conversation analysis is aligned with the program of theory and research known as ethnomethodology. Ethnomethodology examines the "nature and origins of social order. It rejects ‘top-down’ theories that attribute the organization of everyday life to cultural or social structural phenomena understood as standing outside of the flow of ordinary events. Instead it makes use of a 'bottom-up' approach, and seeks to “recover social organization as an emergent achievement that results from the concerted efforts of societal members acting within local situations" (Maynard & Clayman, 2003). Social order does not proceed from a set of stable facts; instead, social action is achieved every day by independent individuals acting in concert. It considers practical reasoning and everyday common sense as an object of study. Specifically, ethnomethodology articulates the “taken-for-granted” rules of order that people use when negotiating everyday matters; the sorts of things that seem obvious to the actual participants in the situation. Ethnomethodology began in the mid part of the twentieth century through the work of sociologists who then began the program of study known as Conversation Analysis.
  • 4. IA Summit 2010 Tanya Rabourn Background & History Harvey Sacks gesture & gaze 4 Initially, naturally occurring conversation was considered by such linguistic scholars as Chomsky to be too messy to be studied in detail. Yet Harvey Sacks, as he listened to recordings of phone calls to a suicide hotline, noticed the social function certain rhetorical moves served and he began to develop an empirical method for investigating conversational sequences. Unlike other fields that studied communication, conversation analysis examines naturally occurring talk instead of an idealized form. Recordings are important because people vary words with pitch rises and falls, etc. The history of conversation analysis is tightly tied with technology. It wasn't until audio and video recording became common that researchers were able to inspect natural conversation closely. Despite the early attention to recordings of phone calls, Conversation Analysis includes both linguistic and paralinguistic features of conversation such as speech, gesture and gaze. It's the study of talk-in-interaction and asserts that talk is used for something, it is functional. Talk performs work. Talk makes things happen and by using conversation analysis, we begin to uncover how.
  • 5. IA Summit 2010 Tanya Rabourn “I see what you did there.” flickr: Ame Otoko 5 One thing I want to stress is that what conversation analysis uncovers is something as capable members of society, who are familiar with our local situations, we demonstrate to each other every day that we already know how talk works. We tacitly orient to these norms. Conversation analysis helps us to make them explicit as we articulate these patterns so we can consider how to better design social media spaces.
  • 6. IA Summit 2010 Tanya Rabourn Repair • Meaning is actively negotiated • Person with “next turn” interprets flickr: Stuck in Customs 6 As you know from your work, it’s usually when something goes wrong that the mechanism for the right path becomes apparent. When things go smoothly, the rules seem invisible, as you can't imagine things happening any other way. So conversation analysts study such things as repairs of conversational misunderstandings – those moments in a conversation when you realize that your interactional partner assumes you meant one thing when you meant another. Or when half way through a conversation your realize you misunderstood a crucial statement she made at the beginning. The work you do to remedy this problem is known as a repair. Importantly this sort of analysis involves only what the participants themselves can see and hear without addressing inner feelings or motivations. The meaning of a conversation is not found in language rules and dictionary definitions or as something that exists within the individuals; instead, it is actively negotiated among participants. It is the person who takes the next turn at talk who determines what interactional meaning the prior statement or “utterance” has. It’s a fundamental CA principle that interactions are structured as sequences in which each move incorporates the participants interpretation of the immediately preceding move (Wasson, 2000).
  • 7. IA Summit 2010 Tanya Rabourn Schegloff, Emanuel A. (1992). Repair After Next Turn: The Last Structurally Provided Defense of Intersubjectivity in Conversation. The American Journal of Sociology, 97(5), 1295-1345. 7 [Sound File] Before we move on to how to apply conversation analysis to design I’m going to spend just a few minutes showing you what the typical output of CA research looks like to make sure we all a good understanding of it. Continuing with the idea of repairing misunderstandings, here is a brief exchange between two sisters: Agnes and Portia on the phone. Portia starts the close of the conversation but it is taken by Agnes as a complaint about her not having time to see her sister. Portia utters the repair-initiating component “Oh” which is followed by an acceptance of Agnes’s excuse for not visiting. Portia: Well, honey? I'll probably see you one of these days Agnes: Oh god yeah, I just couldn’t get down there. Portia: Oh- Oh I know. I'm not asking you to come down. In analyzing this bit of talk we note how Agnes received Portia statement but even more importantly we can see how Agnes reacted to Portia’s reception of her original statement. We can see the unfolding of shared understanding or “intersubjectivity.” So for conversation analysts such a brief conversation is rich with data about how we perform work with talk.
  • 8. IA Summit 2010 Tanya Rabourn Situated & Generalizable “...the locus of order here is not the individual ... nor any broadly formulated societal institution, but rather the procedural infrastructure of interaction, and, in particular, the practices of talking in conversation.” Schegloff, Emanuel A. (1992). Repair After Next Turn: The Last Structurally Provided Defense of Intersubjectivity in Conversation. The American Journal of Sociology, 97(5), 1295-1345. flickr: smcgee 8 Just like the highly descriptive very local ethnographies of particular situations that you might be more familiar with, conversation analysis involves the detailed study of situated local action. But unlike a lot of qualitative research, through observation, conversation analysts attempt “to find patterns that can be formulated as normative behaviors” (Laursen, 2005). They gather multiple instances of the same conversational work in similar situations. These instances are generalizable and that is one of the things that makes this situated research so useful for design.
  • 9. IA Summit 2010 Tanya Rabourn “write like a human” Crumlish & Malone (2009). Designing social interfaces: (1st ed.). “conduct a dialog in as human-like way as possible” Suchman (2007). Human-machine reconfigurations: plans and situated actions (2nd ed.). flickr: mikebaird 9 So, when Christian and Erin, in their book Designing Social Interfaces: suggest we should “write like a human” and HCI researchers recommend that computers should "conduct a dialog in as human like way as possible," conversation analysis offers a body of research that explains what exactly that means.
  • 10. IA Summit 2010 Tanya Rabourn • Computers as partners • Analyzing usability sessions flickr: blakespot 10 Lucy Suchman was one of the researchers who in the mid 80s made use of conversation analysis by considering the computer an interactional partner in a conversation. They used conversational analysis to study the order of conversation between a user and computer (Suchman, 2007). Other HCI researchers have suggested that conversation analysis provides a rigorous method for analyzing usability sessions (Douglas, 1995; Nieminen, Karjalainen, Riihiaho, & Mannonen, 2009). What exactly goes on between the facilitator and the user who is supposed to articulate what they are thinking as they attempt a task? Some researchers suggest that CA gives us a way to analyze this conversation and note the trouble spots.
  • 11. IA Summit 2010 Tanya Rabourn Zawinski's Law of Software Envelopment “Every program attempts to expand until it can read mail. Those programs which cannot so expand are replaced by ones which can.” flickr: c0t0s0d0 11 Zawinski's Law of Software Envelopment “Every program attempts to expand until it can read mail. It’s a snarky comment by programmer Jamie Zawinski about software bloat. But there’s a reason that mail is what these programs eventually incorporate. People don’t want to talk with computers they want to talk with other people.
  • 12. IA Summit 2010 Tanya Rabourn Mediated Communication • computer-mediated communication (CMC) • Internet relay chat (IRC) • Multi-user dungeons (MUDs) • MUD Object Oriented (MOOs) flickr: Jeff Maurone 12 The literature that addresses computer-mediated communication includes a substantial body of work that considers text-based, synchronous computer-mediate communication and language use. In particular, IRC and chat within text-based virtual worlds such as MUDs and MOOs in the 1990s received the attention of researchers. These researchers considered the communicative practices in such environments and noted that they exhibit characteristics of both oral and written communication. (Werry, 1996) Some of you may be familiar with Walter Ong and his assertion that we’re entering a secondary orality made possible through these new means of communication. However, online conversation differs from ordinary conversation in a number of ways. Ordinary conversation relies on the allocation of turns. Online conversation does not have overlaps or interruptions (Werry, 1996, p. 51). Instead, conversational participants construct each utterance in isolation. They send each of them to the system that then posts them in the order received.
  • 13. IA Summit 2010 Tanya Rabourn Mathematical Model Dubberly, Hugh, & Pangaro, Paul (2009). On Modeling: What is conversation, and how can we design for it? interactions, 16(4), 22-28. flickr: misterbisson 13 Previously the mathematical model of communication dominated engineering. And even as recently as this past summer in the ACM magazine, Interactions, the authors suggested that designing for conversation could simply be modeled on Shannon’s model of communication. In this model there is the information source and the signal that must be transmitted through varying levels of noise to the receiver. Although highly useful for developing communications systems, it’s not so useful when designing for conversation. It leads to some seemly practical yet unsatisfactory design decisions. For example, you might think that if your users repeatedly ask other users for some bit of information, then it makes sense to add a feature that makes that information available. However, their asking might be doing other work than just eliciting information. Without thoughtful consideration of what interactional work that question is doing you risk robbing them of that conversational resource. One early enhancement to text-based chat was providing users a way to set up a public profile where they could enter their age, sex and location. However, researchers found that the most common question in the chat continued to be “what is your age, sex or location?” Or sometimes a variation that revealed they had viewed that person’s profile, but they asked are you really in “Florida?” or whatever the user’s location was. Those questions were conversation starters. In designing for conversation, it’s important to observe the interactional work occurring, not just the lexical meaning. Don’t treat conversation as a pure exchange of information.
  • 14. IA Summit 2010 Tanya Rabourn CA & Iterative Design Allison Woodruff, et al at PARC • integrating a conversation analyst in an iterative design process • museum audio guide • make it a social experience flickr: Thomas Hawk 14 There haven’t been many guides written on how to actually integrate conversation analysis into the design process. However, one PARC design team did bring in a conversation analyst to work with them on a project and they’ve published their experience. The project was a descriptive audio system for a museum and they were particularly interested in fostering a social experience for visitors. An initial prototype used a speaker system so the experience could be shared. Afterward, they learned from user interviews that visitors valued both the shared experience with their friends and being able to move at their own pace. So the design challenge was to satisfy two conflicting goals. The conversation analyst was able to analyze the interaction with the first prototype to inform the design of the second. Specifically she learned that visitors allocated the audio a turn making it a conversational partner with themselves. Understanding how visitors to the museum integrated the audio into their personal conversations with the person who accompanied them was key to developing the second more sophisticated prototype that actually allowed visitors to • independently listen to the audio narration • listen in on what their friend was currently listening to • and talk with their friend, even over the audio This next iteration was successful and met the conflicting design goals. It was the nuanced understanding of interaction that allowed them to make appropriate design decisions. They analyzed the organization of talk with the second prototype and noted that users fluidly went back and forth between independent and social experiences continuing to incorporate the audio as if it were an interactional partner in the conversation.
  • 15. IA Summit 2010 Tanya Rabourn “Coherence in the context of computer- mediated communication can be defined as sustained, topic-focused, person-to-person exchanges. Addressivity is a strategy for creating cross-turn coherence online.” Honeycutt, & Herring (2009). Beyond Microblogging flickr: Jeff Maurone 15 In order to establish that a conversation is taking place there must be “coherence.” Coherence is marked by sustained, topic-focused, person-to-person exchanges." These person-to- person exchanges require some way of indicating whom we are addressing. In ordinary conversation, there are multiple ways of doing this including just a simple gaze. In the paper, Beyond Microblogging, the authors discuss user appropriation of Twitter for conversation. They ask "[h]ow well does Twitter support user-to-user exchanges?" In particular they consider the function of the @ sign as a marker of addressivity. In their data they found that there was "short, dyadic exchanges relatively often, along with some longer conversations with multiple participants that are surprisingly coherent. These conversations are facilitated in large measure by use of the @ sign as a marker of addressivity (i.e., to direct a tweet to a specific user) and the ability to “follow” other users, which aid users in tracking conversations."
  • 16. IA Summit 2010 Tanya Rabourn Participation Framework • Multi-party conversation • On Twitter, “surprisingly coherent.” flickr: Ed Yourdon 16 In ordinary conversation, two-person or dyadic conversation is fairly straightforward with speaker and recipient constantly swapping roles back and forth. Multi-party conversation though is particularly interesting where the participants work constantly to build the identities of who is speaker and who is recipient again and again. The participants constantly modify the participation framework. The speaker can designate the next speaker through direct address, or gaze or a number of other actions depending on local practice. So in multiparty conversation a lot of work must be performed to maintain coherence and continually build the participation framework. Shared attention. On Twitter we lack many of these conversational resources which is why the researchers described finding multiparty conversations on Twitter “surprisingly coherent.” In one example from a six person conversation over several turns, the participants used both topicality and addressivity.
  • 17. IA Summit 2010 Tanya Rabourn Participation Framework gsmith: But I was really saying that Twitter didn't borrow it. Twitter users imported it, then Twitter made it part of the system. Paved cowpaths. Rabourn: @saleemkhan @kaleemux @gsmith Tangent: I watched a Twitter user have his 1st IRC experience & his 1st question was "can I use > 140 chars 17 Example 1
  • 18. IA Summit 2010 Tanya Rabourn Participation Framework chrisfahey: Most Twitter clients autolink your @ppl and #tags so the more conversational and contextual you are the more *interactive* your tweets are. chrisfahey: @Rabourn The article credits IRC for @ on like page two. Rabourn: @chrisfahey Doesn't the footnote make it sound like they think it's spreading to places like flickr from twitter? I think other way around. 18 Example 2
  • 19. IA Summit 2010 Tanya Rabourn Footing (Goffman) • Not the author of the utterance only the “animator” • Can be explicit or implicit • Journalists for example may shift footings throughout an interview flickr: Ed Yourdon 19 Shifts in footing can also be a problem with Twitter. Goffman introduced the footing concept in order to explore the nature of involvement and participation in social interaction (Clayman, 1995). Speakers can use interactional resources to show that their words are not entirely their own. For example, a journalist might explicitly say to a politician “your political opponent suggests that you’re soft on crime.” However, journalists also conduct interviews where they articulate a question that suggests a notion not their own without attribution, such as “Mr. President, will the health care bill place an undue burden on small businesses?” In the course of an interview, a journalist may shift footing several times, animating his own words and then articulating those of others’ to display provocative viewpoints. Participants on twitter may echo or paraphrase others’ tweets by attributing them using RT or via and the original author’s username. However, when giving voice to statements outside of Twitter, shifts in footing lead to confusion.
  • 20. IA Summit 2010 Tanya Rabourn Footing Nedaagain: today the dom. win at bazar tehran this is such a success that from now every wed. their will closed down the bazare::) #iranelection #gr88 Nedaagain: bazar is busy is closing bazar is closing one daughter one mother one man were arrested and bazari #iranelection madyar: Ppl like @Nedaagain damage our green revolution with rumors Nedaagain: what rumors you are talking about confirm ?? @Artemis_iaRT @madyar Ppl like @Nedaagain damage our green revolution with rumors #iranelection onlymehdi: @Nedaagain and you were there Nedaagain: @onlymehdi no i was by persian radio& ppl were calling in with news & its very reputable as well as today they confirmed #iranelection 20 An example from the data that makes up the posts to twitter concerning the contested Iranian election this past summer demonstrates this confusion. Just to sum up the event, people in Tehran were calling in to a Persian radio station to report a possible strike in the bazaar, a Twitter user was listening...translating the statements into English and posting them on Twitter. It was over a period of a few hours. Those who came late had missed any explicit attribution for the claims being made and questioned him, accusing him of being the author of these statements and not just the animator.
  • 21. IA Summit 2010 Tanya Rabourn Monitor turns in-progress Churchill, Elizabeth F. (2008). Ps AND Qs: Keep your hair on: designed and emergent interactions for graphical virtual worlds. interactions, 15(3), 38-41. 21 Bob Moore, a game designer at Multiverse, notes that in virtual worlds quasi-synchronous chat lags behind synchronous avatar motion. Moore explains that “… a key feature of real-life conversation is that you can hear a turn unfolding in real time. This enables you to do things like determine who should speak next, anticipate precisely when the turn will end so you can start your next turn with minimal gap and overlap, and even preempt the completion of the current speaker’s turn if you don’t like the direction in which it’s going. In other words, the ability to monitor other people’s turns in-progress is a requirement for tight coordination in conversation. Most virtual worlds use IRC- or IM-style chat systems and therefore do not allow players to achieve this tight coordination among their turns at chat and avatar actions. The result is an interactional experience that feels very unnatural (at first) and that motivates players to invent workarounds to the system.”
  • 22. IA Summit 2010 Tanya Rabourn Summary • Notice “workarounds” • Interactional work, not lexical meaning • Don’t treat conversation as a pure exchange of information flickr: Jeff Maurone 22 Much like ethnography, conversation analysis uncovers opportunities for design. It is a technique for organizing our observations to derive models. So we’ve discussed the history of CA and how it’s traditionally done. We’ve also briefly gone over how to use it in iterative design. I’ll just leave you with three things to think about as you observe conversations in the systems you design. • Notice user “workarounds.” We discuss this a lot, often calling it desire lanes or paving the cow paths. • Observe the interactional work occurring, not just the lexical meaning. • Don’t treat conversation as a pure exchange of information.
  • 23. IA Summit 2010 Tanya Rabourn Questions? flickr: misterbisson 23
  • 24. IA Summit 2010 Tanya Rabourn Recommended Reading Churchill, Elizabeth F. (2008). Ps AND Qs: Keep your hair on: designed and emergent interactions for graphical virtual worlds. interactions, 15(3), 38-41. Dourish, Paul (2004). Where the action is: the foundations of embodied interaction. Cambridge, Mass.: MIT Press. Honeycutt, Courtenay, & Herring, Susan C. (2009). Beyond Microblogging: Conversation and Collaboration via Twitter. Paper presented at the Proceedings of the 42nd Hawaii International Conference on System Sciences. Nieminen, Marko, Karjalainen, Sari, Riihiaho, Sirpa, & Mannonen, Petri (2009). Towards Fine-Grained Usability Testing: New Methodological Directions with Conversation Analysis. In Masaaki Kurosu (Ed.), Human Centered Design (pp. 879-887). Suchman, Lucy (2007). Human-machine reconfigurations: plans and situated actions (2nd ed.). Cambridge ; New York: Cambridge University Press. Woodruff, Allison, & Aoki, Paul M. (2004). Conversation analysis and the user experience. Digital Creativity, 15(4), 232-238. Woodruff, Allison, Szymanski, Margaret H., Grinter, Rebecca E., & Aoki, Paul M. (2002). Practical strategies for integrating a conversation analyst in an iterative design process. Paper presented at the Proceedings of the 4th conference on Designing interactive systems: processes, practices, methods, and techniques. 24
  • 25. IA Summit 2010 Tanya Rabourn Bibliography Churchill, Elizabeth F. (2008). Ps AND Qs: Keep your hair on: designed and emergent interactions for graphical virtual worlds. interactions, 15(3), 38-41. Clayman, Steven E. (1995). Footing in the achievement of neutrality: the case of news interview discourse In Paul Drew (Ed.), Talk at work interaction in institutional settings (pp. 163-198). Cambridge: Cambridge Univ. Press. Crumlish, Christian, & Malone, Erin (2009). Designing social interfaces: [principles, patterns, and practices for improving the user experience] (1st ed.). Beijing; Cambridge: O'Reilly Media. Douglas, Sarah A. (1995). Conversation analysis and human-computer interaction design. In Peter J. Thomas (Ed.), The social and interactional dimensions of human-computer interfaces (pp. 184-203). Cambridge; New York: Cambridge University Press. Dourish, Paul (2004). Where the action is: the foundations of embodied interaction. Cambridge, Mass. [u.a.]: MIT Press. Dubberly, Hugh, & Pangaro, Paul (2009). On Modeling: What is conversation, and how can we design for it? interactions, 16(4), 22-28. Honeycutt, Courtenay, & Herring, Susan C. (2009). Beyond Microblogging: Conversation and Collaboration via Twitter. Paper presented at the Proceedings of the 42nd Hawaii International Conference on System Sciences. Laursen, Ditte (2005). Please reply! The replying norm in adolescent SMS communication The Inside Text (pp. 53-73). Maynard, Douglas W., & Clayman, Steven E. (2003). Ethnomethodology and conversation analysis. In Larry T. Reynolds (Ed.), The Handbook of Symbolic Interactionism (pp. 173-202). Walnut Creek, CA: AltaMira Press. Nieminen, Marko, Karjalainen, Sari, Riihiaho, Sirpa, & Mannonen, Petri (2009). Towards Fine-Grained Usability Testing: New Methodological Directions with Conversation Analysis. In Masaaki Kurosu (Ed.), Human Centered Design (pp. 879-887). Schegloff, Emanuel A. (1992). Repair After Next Turn: The Last Structurally Provided Defense of Intersubjectivity in Conversation. The American Journal of Sociology, 97(5), 1295-1345. Suchman, Lucy (2007). Human-machine reconfigurations: plans and situated actions (2nd ed.). Cambridge ; New York: Cambridge University Press. Woodruff, Allison, & Aoki, Paul M. (2004). Conversation analysis and the user experience. Digital Creativity, 15(4), 232-238. Woodruff, Allison, Szymanski, Margaret H., Grinter, Rebecca E., & Aoki, Paul M. (2002). Practical strategies for integrating a conversation analyst in an iterative design process. Paper presented at the Proceedings of the 4th conference on Designing interactive systems: processes, practices, methods, and techniques. (and Flickr users: camera_obscura, sveinnbirkir, nasrulekram, Ame Otoko, Stuck in Customs, smcgee, mikebaird, blakespot, c0t0s0d0, Jeff Maurone, misterbisson, Thomas Hawk, Ed Yourdon, ) 25