communication and collaboration models Look at several levels - PDF document

CSCW Issues and Theory All com puter system s have group im pact chapter 14 – not just groupware Ignoring this leads to the failure of system s communication and collaboration models Look at several levels – minutiae to large scale context: – face-to-face com m unication – conversation – text based com m unication – group working Face-to-face communication Transfer effects • carry expectations into electronic media … • Most prim itive and m ost subtle form of … som etim es with disastrous results com m unication • m ay interpret failure as rudeness of colleague • Often seen as the paradigm for computer mediated communication? e.g. personal space – video m ay destroy m utual im pression of distance – happily the ` glass wall' effect helps Eye contact Gestures and body language • to convey interest and establish social • m uch of our communication is through our presence bodies • gesture (and eye gaze) used for deictic reference • video may spoil direct eye contact (see video tunnel, chap 19) • head and shoulders video loses this • but poor quality video better than audio only So … close focus for eye contact … … or wide focus for body language? 1

Back channels Back channels (ctd) Alison: Do you fancy that film … err 1 … • Back channels include: ` The Green' um 2 … – nods and grimaces it starts at eight. – shrugs of the shoulders Brian: Great! – grunts and raised eyebrows • Not just the words! • Utterance begins vague … • Back channel responses from Brian at 1 and 2 … then sharpens up just enough – quizzical at 1 – affirm ative at 2 Back channels -media effects Back channels and turn-taking in a m eeting … Restricting media restricts back channels – speaker offers the floor (fraction of a second gap) – listener requests the floor video – loss of body language (facial expression, sm all noise) audio – loss of facial expression Grunts, ‘ um ’s and ‘ ah ’s, can be used by the: half duplex – lose most voice back-channel – listener to claim the floor responses – speaker to hold the floor text based – nothing left! … but often too quiet for half-duplex channels e.g. Trans-continental conferences – special problem – lag can exceed the turn taking gap … leads to a monologue! Basic conversational structure Adjacency pairs Alison: Do you fancy that film Sim plest structure – adjacency pair Brian: the uh ( 500 ms ) with the black cat Adjacency pairs m ay nest: ‘The Green whatsit’ Brian: Do you want some gateau? Alison: yeah, go at uh … Alison: is it very fattening? ( looks at watch – 1.2 s ) … 20 to? Brian: yes, very Brian: sure Alison: and lots of chocolate? Brian: masses Sm allest unit is the utterance Alison: I'll have a big slice then. Structure is: B-x, A-y, B-y, A-z, B-z, A-x – inner pairs often for clarification Turn taking � utterances usually alternate … … but, try analysing the first transcript in detail! 2

Context in conversation Referring to things – deixis Utterances are highly am biguous Often contextual utterances involve indexicals: that , t his , he , she , it We use context to disam biguate: these m ay be used for internal or external context Brian: ( points ) that post is leaning a bit Alison: that's the one you put in Also descriptive phrases m ay be used: – external: ‘ t he corner post is leaning a bit’ Two types of context: – internal: ‘ t he post you m entioned’ • external context – reference to the environm ent e.g., Brian's ‘ that ’ – the thing pointed to deictic reference • internal context – reference to previous conversation I n face-to-face conversation can point e.g., Alison's ‘ that ’ – the last thing spoken of Common Ground Focus and topic Resolving context depends on m eaning Context resolved relative to current dialogue focus � participants m ust share m eaning so m ust have shared knowledge Alison: Oh, look at your roses : : : Brian: mmm, but I've had trouble with greenfly. Conversation constantly negotiates m eaning Alison: they're the symbol of the English summer. … a process called grounding : Brian: greenfly? Alison: So, you turn right beside the river. Alison: no roses silly! Brian: past the pub. Alison: yeah … Tracing topics is one way to analyse conversation. – Alison begins – t opic is roses Each utterance is assum ed to be: – Brian shifts topic to greenfly relevant – furthers the current topic – Alison misses shift in focus … breakdown helpful – comprehensible to listener Speech act theory Breakdown Breakdown happens at all levels: A specific form of conversational analysis t opic, indexicals, gesture Utterances characterised by what they do … … they are acts Breakdowns are frequent, but e.g. ‘ I 'm hungry’ – redundancy m akes detection easy – propositional meaning – hunger ( Brian cannot interpret ‘ they're … sum m er’ ) – intended effect – ‘ get me some food’ – people very good at repair ( Brain and Alison quickly restore shared focus) Basic conversational act the illocutionary point: – promises, requests, declarations, … Electronic media may lose some redundancy Speech acts need not be spoken � breakdown more severe e.g. silence often interpreted as acceptance … 3

Conversations for action (CfA) Patterns of acts & Coordinator • Generic patterns of acts can be identified • Conversation for action (CfA) regarded as central • Basis for groupware tool Coordinator – structured em ail system – users m ust fit within CfA structure Circles represent ‘states’ in the conversation – not liked by users! Arcs represent utterances (speech acts) CfA in action Text-based communication • Sim plest route 1–5: Most com m on m edia for asynchronous groupware exceptions: voice m ail, answer-phones Alison: have you got the market survey on chocolate mousse? request Fam iliar m edium , sim ilar to paper letters Brian: sure promise but, electronic text m ay act as speech substitute! Brian: there you are assert Alison: thanks declare Types of electronic text: – discrete directed messages, no structure • More com plex routes possible, e.g., 1–2–6–3 … – linear messages added (in temporal order) – non-linear hypertext linkages Alison: have you got … request – spatial two dimensional arrangement Brian: I've only got the summary figures counter Alison: that'll do accept I n addition, linkages m ay exist to other artefacts Problems with text example – ‘Conferencer’ No facial expression or body language � weak back channels So, difficult to convey: affective state – happy, sad, … illocutionary force – urgent, important, … Participants com pensate: ‘flam ing’ and sm ilies ; -) : -( : -) linear conversation area – LHS RHS – spatial simulated pinboard 4

Conferencer (ctd) Grounding constraints Establishing common ground depends on Note separate ‘composition box’ Pin board has similar granularity – transcript only updated grounding constraints ‘cards’ only appear on other when contribution ‘sent’ participants’ screens when – granularity is the contribution edit/ creation is confirmed cotem porality – instant feedthrough simultaneity – speaking together Note separate ‘composition box’ sequence – utterances ordered – transcript only updated Pin board has similar granularity when contribution ‘sent’ ‘cards’ only appear on other – granularity is the contribution participants’ screens when Often weaker in text based communication edit/ creation is confirmed e.g., loss of sequence in linear text loss of sequence Maintaining context Network delays or coarse granularity � overlap Recall context was essential for disam biguation 1. Bethan: how many should be in the group? Text loses external context, hence deixis 2. Row ena: maybe this could be one of the 4 strongest reasons 3. Row ena: please clarify what you mean ( but, linking to shared objects can help) 4. Bethan: I agree 5. Row ena: hang on 6. Row ena: Bethan what did you mean? 1. Alison: Brian's got som e lovely roses 2. Brian: I 'm afraid they're covered in greenfly Message pairs 1&2 and 3&4 com posed sim ultaneously – lack of com m on experience 3. Clarise: I 've seen them , they're beautiful Rowena: 2 1 3 4 5 6 Bethan: 1 2 4 3 5 6 Both (2) and (3) respond to (1) N.B. breakdown of turn-taking due to poor back channels … but transcript suggests greenfly are beautiful! Non-linear conversation Pace and granularity 1. Alison: Pace of conversation – the rate of turn taking Brian’s got some face-to-face – every few seconds lovely roses telephone – half a m inute em ail – hours or days 2. Brian: 3. Clarise: face-to-face conversation is highly interactive I’m afraid they’re I’ve seen them – initial utterance is vague covered in greenfly they’re beautiful – feedback gives cues for com prehension lower pace � less feedback hypertext-based or 4. Clarise: � less interactive have you tried threaded-m essage system s companion planting? m aintain ‘parallel’ conversations 5