Conversation Space: Visualising Multi-threaded Conversation ...

39 downloads 7337 Views 616KB Size Report
as email and newsgroups. Design of such software has been guided by an understanding of ATBC as essentially a text communication (textual metaphor).
Conversation Space: Visualising Multi-threaded Conversation Dimitri Popolov

Michael Callaghan

Paul Luker

Bauman State Technical University 2 nd Baumanskaya St. 5 107005 Moscow R u s s i a +7-(095)-164-9638

De Montfort University T h e Gateway Leicester LE1 9BH, UK +44-(0116)-255-1551

De Montfort UniversiE/ The Gateway Leicester LE1 9BH, UK +44-(0116)-255-1551

[email protected]

[email protected]

[email protected]

conceive of ATBC. We argue that this medium possesses a unique combination of spontaneity of oral and persistence of textual discourse. One outcome of this is a so-called multithreaded discourse structure, with several discourse threads developing simultaneously. However, design of software tools in this medium was guided by the textual metaphor, overshadowing oral properties of the medium and thus narrowing the scope of possible designs.

ABSTRACT This paper explicates the metaphors used to conceive of asynchronous text-based communication (ATBC) software, such as email and newsgroups. Design of such software has been guided by an understanding of ATBC as essentially a text communication (textual metaphor). However, this mode of discourse has many similarities with oral communication as well. The interaction of oral and textual aspects in ATBC gives rise to a phenomenon of multithreaded discourse, where several discourse threads develop simultaneously, which is a unique property of this medium. Our main tenet here is that application of textual metaphor has narrowed the scope of possible designs. We propose a design approach, which explicitly promotes the metaphor of oral communication (conversation) and oral traits of ATBC discourse, while also supporting the multithreaded discourse structure. The consequent interface design challenge is that of creating a way to visualise human conversation that would preserve the spontaneity of oral conversation whilst also utilising the persistent nature of text. This goal has been accomplished by spatial representation of multi-threaded discourse in a shared workspace. Based on this proposed way of visualisation, a prototype tool called 'Conversation Space' (ConverSpace) has been created.

Human discourse is an ongoing social activity rather than tangible material artefact, so visualising it in computer medium is a significant challenge. Thus the visualisation challenge was that of creating a way to represent multi-threaded conversation that would preserve the spontaneity of oral conversation whilst also utilising the persistent nature of text. This methodological analysis was tested and refined by design and implementation of a software tool to support human communication in this medium, which explicitly pr~,motes the metaphor of oral communication - conversation.

2. Oral, textual and asynchronous text-based communication J The theoretical analysis has started with the awareness of the power of metaphors to implicitly guide thinking [7] arid thus the need to make explicate them. In the area of interface design, research on metaphors deals mainly with metaphors of graphical user interface design, e.g. [2, 9, 10]. However, in this investigation we attempt to trace deeper assumptions about human communication in this medium and explicate implications of oral and textual metaphors to better understand how they limit the design and consequently provide better designs.

Keywords asynchronous text-based communication, discourse structure, computer-mediated communication, persistent conversation, multi-threaded discourse.

2.1 Textual metaphor

1. INTRODUCTION

The textual metaphor has permeated virtually every aspect of ATBC software design. It started on the level of titles of technologies, e.g. electronic mail, message or note exchange, hypertext, etc. In the interface features everything has been revolving around icons with tiny envelopes, mailboxes, etc. On a deeper level the communication was conceived as exchange of messages, similar to mail exchange of letters.

With the development of virtual communities over the Intemet, the issues of creating software for support of human communication in distributed working and learning groups are becoming increasingly important. This investigation deals with visualisation of human discourse in asynchronous text-based communication medium (ATBC). Main examples of this mode of interaction are email and newsgroups, but it is also widely used in many group decision support systems, distance education software, etc. We analyse metaphors of oral and textual communication used by designers and researchers to

The most important implications of the text metaphor are connected with the way written language is visualised. There exists a strong tradition of conceptualising and co3asequently visualising printed text. It is usually conceived of and presented as a linear, sequential medium. Printing tradition enforces understanding of text as a sequence of letters forming words,

Permission to make digital or hand copie of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage, and that copies bear this notice and the full citation on the first page. To copy otherwise, to republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee. AV12000, Palermo, Italy. © 2000 ACM 1-58113-252-2/00/0005..$5.00 246

traditional textual layout forces this structure into the linear sequence (Figure 1).

which themselves are combined into linear sentences, etc. Sequences of chapters, pages, volumes are all examples of linear presentation of written language.

Original message:

However, in various branches of literature there can be found many examples of dissatisfaction with this linear pattern and desire to break it to express finer details of meaning or human thought processes. In a scientific literature usage of footnotes and cross-references indicates this. More radical attempts to free the text from linearity are represented by Derrida's Glas [3], in which several threads of textual discourse run in parallel.

I propose to invite Jane on Friday. She will be able to have a good weekend's rest before starting her work. The University will benefit from her contributions.

Reply: >I propose to invite Jane on Friday. I have only been able to make a reservation since Saturday, but I guess it's still OK.

The restrictive nature of traditional print has become especially evident with the explosive growth of hypertext technologies, which support and promote alternative non-linear representations of text. There are works in spatial hypertext [8], showing that spatial dimensions of text can be used to convey additional meaning. Such works allow viewing traditional book layout as a simple instance of spatial hypertext, where for example footnotes are placed at the bottom of the page, and this position is used to convey an additional meaning of remark, namely of its lower importance compared to the main text.

>The University will benefit from her contributions. I agree completely. We just have to make sure her lectures fit nicely into our overall programme.

Figure 1. Email exchange with multiple threads. This structure is evident even in the dyadic conversations, but this appearance of discourse trees becomes most prominent in multi-party dialogues, where there often appear several levels of quoting.

2.2 Oral metaphor

Another new property of dialogues in this medium is their temporal dynamics. Different conversational threads have different paces of development. Also the history of their development is observable, if needed. Combined with the absence of temporal constraints on the discussion, it leads to the feeling that in [6] was labelled as 'rolling present'. Discussion goes on and on as new threads keep on branching.

Precisely because metaphors guide thinking, it was not easy to approach text-based communication as a phenomenon similar to oral one. However, evidence was accumulating that was demonstrating that discourse in electronic media is a new and unique linguistic phenomenon that exhibits properties of both oral and written communication. In the area of ATBC it has been found that this medium gives rise to a new type (register) of discourse that possesses traits of both written and oral discourse [5]. Email allows for a very wide range of communication strategies, but inherently it prompts for fast replies, which adds to the conversational aspects of it. [4]. Rapid message turnaround leads to an increased sense of the "presence" of the audience. This increased sense of presence over more traditional forms of writing is what makes communication in ATBC closer to the oral discourse, affecting its style and structural patterns.

3. Visualisation of multi-threaded conversation In an attempt to promote the desired properties of the dialogue it was proposed to visualise discourse threads on a 2D surface with means of direct manipulation with conversation. The metaphor of oral interaction was consciously promoted, so the discourse was called conversation and contributions were named 'remarks' rather than messages. The main aim of the tool was to achieve combining persistency of text and spontaneity of oral conversation. All this leads to the title of the prototype: Conversation Space (ConverSpace) - spatial representation of computer-mediated conversation.

The awareness of the complex, neither oral nor written, dynamics of ATBC discourse existed at a theoretical level for a relatively long time [1], but was not explicitly used for the design of software tools for support of human communication and collaborative work in this medium. Textual metaphor was so influential in the thinking of designers, that only from their experience, on an intuitive basis, was support of oral aspects of such communication and conversation threads, implemented in many commercial email and conferencing programs, as well as research prototypes, e.g. in [11].

Since this is a new linguistic phenomenon and there are many issues of which we do not have adequate theoretical knowledge, an iterative prototyping approach was chosen to refine the method of visualisation.

3.1 The first prototype The first goal of the layout issues was to free the conversation from the assumptions associated with the print. We wanted to pose as few limitations as possible on spatial layout, bringing the prototype close to the shared whiteboard applications, but with conversational fragments linking by quoting. A user could highlight a text fragment in an existing remark or the whole remark and add a reply remark, which would automatically be linked to the first one. Two shapes of remarks were supported: rectangular and diamond (Figure 2).

2.3 Interaction of oral and textual properties In ATBC discourse users often use quotations to preserve the conversational context of the interaction. This phenomenon illustrates very well the unique nature of the ATBC discourse. Quoting, which is a typical written strategy, is used to promote the oral, conversational feel of the discourse [4]. The usage of quoting, as well as replying to specific messages, gives rise to the so-called multithreaded dialogue, when several topics or threads of dialogue are being developed simultaneously. But

247

Remarks are boxes containi ng some text. ~%a-,::: N:~ ttcally lt,%~ Objects can be linked to other objects' bou ndaries.

Objects can be linked to text se lections ..~

J

Text can contain [ hyperlinks to truly [ integrate Conver ]

I



)nterlacing text selections are shown in darker color

"\1 Textthat does

.]

] notjit into the box is marked

Figure 2. ConverSpace ver. 1. The experience with the first prototype showed that the idea of representing conversation spatially and conversational fragment linking itself was viable, but several problems were also identified.

obscured by the free-form spatial layout, with which it was difficult to visually identify a linear sequence from remarks scattered all over the drawing surface. The adopted form of visualisation was too flexible and as such could not actually support the desired mode of interaction. So now the task was to identify an appropriate spatial structure and to ensure the system actively supports it.

Firstly, having users choose remark type and manually place remarks on the drawing surface created an additional cognitive load for them. This reduced spontaneity of response and made the prototype closer to such activities as group hypertext writing with richer means of expression, but standing closer to the writing pole of the discourse than the oral. This flee-form layout posed problems in navigation and the location of new contributions, because from the technical properties of the system users could not predict where the remark would be placed by another user.

3.2 The second prototype This experience has lead to the second version of the prototype. The conversation was still represented spatially, on a 2D surface but now the system maintained the two axes, and remarks were positioned automatically according to them. The horizontal axis represented time. The vertical axis represents topical structure. The combination of these two axes resulted in a tabular layout (Figure 3).

Secondly, in spite of the 'rolling present' nature of time in ATBC, every single thread was developing in a separate, but still linear time. In traditional software, the time dimension is usually visualised as a list of messages sorted by time. This aspect was lost in the prototype, because now both spatial dimensions were used.

The prototype was distributed to several specialists in IT to gather feedback and suggestions on its potential improvements. These test trials has showed that the spontaneity of discourse was restored. In fact, the process of remarks composition felt so spontaneous, that the explicitly asynchronous nature of the prototype started to feel restrictive, Consistent rules for placement of remarks allowed users to concentrate on its contents and thus increased the speed of interaction. There were identified two main potential uses for the prototype: (a) as an informal conversation medium and (b) as a medium for collaborative design, because persistent and structured transcript made possible easy capturing and review of design rationale.

Thirdly, the threads in the on-line discourse are more flexible than could be represented in the prototype, because users often reference several messages in their replies, so branches can not only separate, but also merge. This requires a structure that is not exactly tree-like, but rather more in the form of a network. Also, there was a realisation that reading and navigating this threaded structure is still a linear process, because every separate thread has its own pace and temporal dynamics. This was

248

Some thoughts on the tool and its evaluation.., l 1) It's a gmupware tool- It I--

llsnafalrtotestema, "

/

~ t h e moment i am . . .[ : ' . . . interested in ] I comparing ~

[~"~ ~ I

:

I~1

~ut surely newsgroups ~ave an element of 'sh~red space ~ abOUt them, but are just =OEe linear

~1

~,~J

I

;i Fell me again how multi-user the tool $ .... e/hen I'm using a ~ace no-one else,

Yes, it is explicitly meant to be ~lsynchronous. 3o wher'l you are oggedinto a ,

]~.eEE a~e also questlons of supportlng f r a ~ e n t s li~b~J~ig.[

:o::,U ¸,'¸¸¸¸} But SLIreiy Inewsgroups ihave an element

Figure 3. ConverSpaee ver. 2. 4. C O N C L U S I O N S This paper stresses importance of challenging assumptions underlying media and suggests metaphor analysis as a way to identify potential for new software designs.

[4] Eklundh, K.S. and Macdonald, C. The use of quoting to preserve context in electronic mail dialogues. IEEE Transactions on Professional Communication, 37(4) (1994), 197-202

•Human discourse is a very fluid and adaptable phenomenon. Our theoretical understanding of its 'real-world' dynamics, especially as connected with software interface features, is very tentative. It is hoped that findings obtained from work with ConverSpace prototypes could be generalised to other forms of electronic discourse.

[5] Ferrara, K., Brunner, H. and Whittemore, G. Interactive written discourse as an emergent register. Written Communication, 8(1) (1991), 8-34. [6] Hiltz, R. and Turoff, M. The Network Nation: Human Communications via Computer. The MIT Press, Cambridge, MA, 1993

Further work includes development of a production-quality software based on the ConverSpace prototype and its evaluation in 'real-life' conditions. It is planned to use this system as a part of a virtual learning environment within the framework of a joint project in distance education between De Montfort University and Bauman Moscow State Technical University. All technical information and detailed interface description are freely available from authors.

[7] Lakoff, G. and Johnson, M. Metaphors We Live by. Chicago: University of Chicago Press, 1980 [8] Marshall, C.C., Shipman III, F.M. and Coombs, J.H. VIKI: spatial hypertext supporting emergent structure. In proceedings of ECHT'94, 13-23. [9] Norman, K.L. and Chin, J.P. The menu metaphor: food for thought. Behaviour & Information Technology, 8(2) (March-April 1989), 125-134.

5. REFERENCES [1] Black, S.D., Levin, J.A., Mehan, H. and Quinn, C.N. Real and non-real time interaction: unravelling multiple threads of discourse. Discourse Processes, 6 (1983), 59-75

[10]Sorensen, E. Metaphors and design of human interface. In A.R.Kaye (ed.) Collaborative Learning through Computer Conferencing. Berlin: Springer Verlag, 189-199, 1982.

[2] Colgan, L., Spence, R. and Rankin, F. The Cockpit Metaphor. Behaviour & Information Technology, 14 (July-August 1995), 251-263.

[l l ] Whittaker, S., Swanson, J., Kucan, J. and Sidner, C. TeleNotes: managing lightweight interactions in the desk-top. ACM Transactions on Computer-Human Interaction 4(2), (1997), 137-16

[3] Derrida, J. Glas. University of Nebraska Press, Lincoln, NE, 1986

249