Data Sessions

From IMC wiki
Jump to: navigation, search

Sam - 08/05/2013

Herb Clark – grounding similar to the gaze (nonverbal) escalating to verbal

Laveilt (sp?) – when restarting where do you start from?

Collaborative playing – teacher repairing the student

Is the teachers nonverbal similar to if they were in the role of secondary recipient during three-way interaction – stopping their nonverbal feedback?

Text based speech - Where the interruption happens predicts how far back you go back when you repair.

Previous experience of being told to continue makes you less likely to request feedback later… importance of context – Herb Clark.

Comparison with clinical data laura – nodding to say keep talking – keep going.

Links with topic shift – Kendon: Moving out and in again at the end /change of topic.

Laura - 06/02/2013

CA transcriptions of 210 declarative questions

  • More use of them
    1. make doctors rate therapeutic relationships better and
    2. make patients rate it worse
  • Tend to be of two types
    • Repeats (partial/exact?
    • Inferences
      • subset of these are so prefaced...
  • These tend to be interpretations
    • Often about how the patient is feeling
    • Even if that's not how they have been talking about it
  • Shifting the focus - like Saul's stuff (below)
  • 'So' can be used to close off a sequence
  • These seem to be routine summaries (diagnostic)
  • Parallels with:
  • Responses (acceptance of challenge of reinterpretation)

How to justify selection criteria (or whether to at all!)

  • Example CP213
    • Pauses with looking at notes before so prefaced declaratives
    • Dr writes notes and so is a continuation of previous stuff (bridge across the gap)
    • Status of pause (as a non-verbal turn)
    • Notes as a conversational device not as anything useful in the future
  • Example AP011
    • 'episodes' characteristic of diagnostic
    • declaratives as chunking devices
      • or directing attention to some feature of the prior chunk of talk
      • pulling together a bunch of stuff into a coherent narrative
    • Note-taking behaviour of doctor is similar - taking notes before summary
    • But no overt pause because patient is talking through it

These examples seem to have different characteristics but also similarities:

  • Structuring devices
  • Summarising prior talk
  • Differences seem to be about how the patient responds not about what the doctor is doing

Extra question: lexical overlap (do Drs pick up on patient language or vice versa)

Saul - 23/01/2013

Data looking for aesthetic reassessment/disagreement subdialogues - searched BNC for `beautiful'

  • Specific conversation: between Katherine, customer and Stefan, a dealer??
    • Is Katherine another dealer or a customer with a gallery? This would affect value judgements etc
  • Context (assumed) flicking through a catalogue
  • Segments to pick out areas of disagreement - assessment attribution segments
    • Clarification sequences (loads of them - may be a non-confrontational way to indicate disagreement)(70-78)
    • Criteria (authorship interchangeable with assessments like beautiful etc)
  • Different factors for positive/negative assessments
    • 78-92 switch which are the 'relevant' (salient?) factors depending on their assessments
      • Relevance theory would have no way of dealing with this I don't think (the usual probem of how do we resolve what is relevant at any given point in a conversation; here it seems to be different for each interlocutor)
  • Markedness or otherwise of pauses
    • Depends on whether they are within an assessment sub-dialogue or between
    • Have the catalogue as a shared locus of attention (and indicate shift of focus, like f-formation shifts)
    • Contemplative pauses between segments not as marked? Topic shifting device
  • Use of deictics (this/that etc)
  • Signature segment switches referent at line 153
  • Conversation about Jackson Pollack - engineers
    • Jackson Pollack is a trained monkey!
    • Relevance of context of pauses (could be conversationally relevant non-verbal turns - can't tell by audio)
    • Also context of what is shared (what is common ground?)
    • Tanya Strivers: Transformative Answers

07/11/2012 - Sam

Musically 'turns' analogous to conversational

Teacher using space to demonstrate whether student should continue playing

  • Repair if tutor interrupts student / mistakes needing correction
    • Different if student is aware of mistake
      • Non-awareness
      • Full awareness
      • Awareness that the tutor thinks they have done something wrong, but not awareness of what
    • Interacts with awareness of tutor
  • Are there cases where something goes wrong, but it is ignored/not repaired
  • Types of error (are they immediately correctable or not?)
  • Transition Relevance Places in music (ends of phrases) - some interruptive, some not
  • Negotiations when there is a mistake
  • Repair in conversation: independent of error
    • no a priori truth/falsity of what has been said (in contrast with music?)
    • repair is dependent on what they choose to treat as needing repair
  • Do analogies hold?

Example fragments - 1:

  • Annotate in-breaths for turns
    • Breathing in synchrony... Mirror neurons...

Example fragment - 2:

  • Shorter referent (as known - this could be another analogy with conversation)
  • Restarts of music not obvious to non-musical people - needs to be obvious where they are
    • Way of displaying it
      • Music therapy micro-analysis might help
      • CA in different languages?
  • Playing things as substituting of speech
    • Several mistakes in passage;
      • What is being picked up on as repairable (Sacks: everything of all types (morphology, syntax, semantics, etc) is potentially repairable)
  • Self-repair of phrase - (is it other-initiated (i.e. in response to something the teacher does) or self-initiated
  • Feedback: backchannels (movement when error detected, but like a continuer feedback, not one of the moving in wanting to take the turn variety)
    • Student leaning forward; flagging something
  • Competing start (i.e. both trying to take the turn at a TRP)

17/10/2012 - Saul

Using updated toolkit of CA things fro Pomerantz, 1984

Data from Tino Sehgal installation at Tate

  • Recordings inaudible - set-up not quite good enough, too much background noise etc.
    • Need better mike
    • Filtering of bacground noise using Audacity
  • Approval questions (pilot?)
    • Informed consent etc - ethics issues
  • Time/effort
    • How much data needed - about collecting aesthetic conversations
  • Site: Balcony; could another site be the cafe (good contrast - but would they talk about the Sehgal)
    • People on balcony have positioned themselves outside of artwork
    • Moving
  • People talking about art don't do it in front of it

Extract BOB and SUE

  • Pauses
    • How marked are they when they are looking at something else
    • Cues available?
    • Section where Bob has extensive pauses within own turn
  • Using visual spectacle as a device
    • For topic shiftng
    • Turn-taking

Extract BNC

  • Looking at an art catalogue
  • What you know and how you know it
    • Also, line 20, she is asserting knowledge even though is wrong (knowledge of Klee)
  • His subjective strong assessment (33) undermined by her - but not taking it head on
    • i.e. shift of criterion
    • His assertion about how it looks, she undermines it with question about whether it sells
  • Interesting sequence of repeats / questions / trouble; 48-66
  • Sequences of clarification requests (again at 68-75): very marked
    • Rights to epistemic knowledge - who has rights to claim what is known etc

10/10/2012 - Laura

Questions: looking at declaratives.

AP045: Medical student present so could affect the dialogue

  • Contrasting perspectives - doctor uses repetition and contrastive markers to explicitly formulate the perspective the patient has compared to what she 'should' have
    • e.g. you think you weigh far too much... but what do other people say
  • When asking wh-Qs, repairs - so softening it? Or changing the structure
  • Seems to be shifting the focus (subjective-objective?) - like Saul's parameter shifting
    • her perspective/something more evidential
  • enumeration?
  • Gaze and gesture?
    • Including notes as an interactional device (what is he actually writing?)
    • Could be annotated for when he takes notes
  • Dr receipts (including assessments of what patient says)
  • Inviting agreement with a candidate understanding - which is successful
    • Positive declaratives designed for a 'yes' (cf. Heritage papers on question design)
    • Y/N less strong than declaratives and tags stronger still
      • Are you happy?
      • You are happy?
      • You are happy, aren't you?
  • Interesting tussle about meals (tea/supper) lines 13-20
  • Different consequences of different receipt tokens (yeah; repetition; well, xxx; etc)
  • Verbatim repeats from doctor seem to be demonstrating lack of understanding
    • Responses seem to be more elaborate in these cases?
  • Summarising ones seem to be showing understanding
    • i.e. using declaratives for opposite purposes but differently designed
  • Seems to be a lot of work in formulating the questions
  • 45-51 interesting sequence about weight
    • P doesn't respond as much as desired/expected
      • pauses; avoidance; minimal responses
    • Dr re-raises her initial assessment almost word for word (twice)
    • 'problem' is way she feels not actuality
  • Interesting interpretation from psychiatrists - felt doctor was distant/abstract (could've said he felt she looked fin, not abstract others)
    • Feels that Dr is "collecting data" from P, not having an interaction

AP045 2.15-3.49

  • Many of the Dr's statements are inviting agreement with a hypothesis - if they were problematic they'd be getting accounts.
  • line 2 correction: going -> doing.
  • line 10: interesting use/thematising of "perspective" here - is this a keyword used in Dr's training? It has that feel, and would be interesting to know to what extent this kind of perspective-talk is trained.
  • line 19 correction: right right right -> Oh right right
  • on lines 19, 29, 42, 48, 62, and 67 the Dr consistently begins a question seeming to be closed/y/n and self-repairs to modify it up to a wh question - in fact this happens before every wh question.

25/09/2012 - Mary


  • Looking for a global way of identifying agreement/diagreement - is this different for groups with patient?
  • Does patient do disagreement differently?
    • Lack of flexibility of patient?

example transcript 17

  • Possible strong disagreement marker - looking outside of the O-space?
  • Looking down?
  • Failed transitions following acknowledgement tokens (Jefferson)
  • Head roll/rocking - moderating opinions? (line 157)
    • Early refusals to engage with patient's topic - line 4-5 and 12-13
  • Different orientations to the task?
    • C imagining scenario - task orientation to finish? Definiteness of things she says.
      • Patient may be used to being asked about things so have a set way they think they ought to respond
    • A/B trying to perpetuate the task (discussing in the abstract)?
  • Line 97 patient repeats (agreeing) and then extends (disagreeing) - i.e. correcting
  • Line 127 responding to an out of sequence turn from patient
  • Patient not engaging or not being allowed to engage?
    • Has she already made up her mind or is that just what the others assume?
    • Patients either not wanting to make a decision at all or making it straightaway and not changing?

How much do they talk about what they aren't supposed to be talking about? (deflecting)

  • Who gives an opinion?
    • Patient instantly - and sticks to it?
    • Strength of opinion? (patient seems to not use evidentials?)
      • Saying who should stay also less strong than who should go.
  • Nodding and head orientation - combination relevant in agreement
    • Patient in disagreement pulls head away

Differences in control group? Group 27

  • Code turns that are obviously
    • Opinion
    • Agreement
    • Disagreement
  • See if there is anything generalisable from them

12/09/2012 - Laura

Formulating Treatment Decisions in Schizophrenia

  • How do clinicians recommend treatment changes - and how do patients respond
  • Comparison with primary care data
  • mainly suggestions? (intuitvely)

Example extract CP221 - 3.41-4.44

  • Dosage increase
  • Offer
  • pronoun "I can increase..."
  • Outright rejection by patient
  • Earlier dialogue is doctor establishing how much responsibility the patient can have?
  • Subsequent negotiations:
    • "we"
    • disagreement
    • patient wants medication - but not the one the doctor wants to increase

Example extract CP250 14.32-15.15

  • Dosage decrease (are these different from drug increases?)
  • Proposal
  • pronoun "we can see if..."
  • accepted by the patient without discussion - but not "readily"
    • seems to be resigned to the decision
    • shrugs? etc

22/08/2012 - Chris/Jacopo

Jacopo study

  • Compare patients’ symptoms by carer present vs’ no carer present
  • What do they say when they are present? What is their role? What are they doing?
  • Is there a difference between carer’s present and not in outcome data? possibly after adjusting for symptoms?
  • Things to look at:
    • Identifying topics – manual coding vs automatic coding
    • Who’s talking about what? -divide topics by participant type
    • Is there a link with symptoms?
    • Is there a link between topics and outcome - specifically looking at the PEQ
  • Other things to think about in future:
    • Audio features frequency/pitch. Not possible to align this with the words, but can just look at overall features

20/06/2012 - Rose/Mary

Balloon task dialogue fragments case studies.

Starting point: How people use body movement to address involvement problems, and get each other to attend/participate (Heath, 1986)

Non-verbal cues for participation in multiparty interaction -- patients with schizophrenia are supposed to have problems with non-verbal cues (from lab-based) -- does this hold in genuine conversations etc.

  • Healthy participants orienting to patient's low level of involvement as a speaker
    • Hierarchy of cues
      1. non-verbal (e.g. glance)
      2. nonverbally offered opportunity to speak at TRP, but declines
      3. explicit "what do you think" (awkward -- put on spot)

Group 17; A,B = healthy participants; P = patient. 2.03 minutes in

  • B looks at A until line 5; then down at line 6 (awkward? scratches leg; or device to look towards patient) and at patient at line 10 (after A says emmm)
  • Who is patient looking at?
  • A looks towards patient but then starts talking when B looks up; not addressing comment to either
  • B mid-turn pause (6) but a potential TRP
    • Both HPs seem to be orienting towards P as a possible next speaker
  • Hypothesis: Other person (non-speaker) also looks at patient more?
  • Is it worth annotating smiling? (c.f. laughter comments from data session 30/5/2012)
  • Line 18; TRP where patient is potentially expected to contribute - A and B both fill (19/21)
  • A looks towards P during 17/18; P shifts backwards
  • B looks don when starts to speak in 23 (after awkward bit) then orients towards A
  • P looks up after A overlaps B and takes turn (after save in 26); looks down before turn is finished (when B comes back in in overlap)
  • Explicit cue (not until 64) comes as the culmination of lots of failed stuff (e.g. line 10; 18; 28)
  • Potential differences between control conversations and ones with patient (lexical alignment; fillers; pauses)
    • CCs - these two (who had fantastic rapport) do it a lot -- also lots of overlap between them (proving my point that it's not always competitive, but can be cooperative!?)
  • Gaze so important; technological issues
  • Coding for gaze done on a frame by frame basis; but approximate

20/06/2012 - Mary

Identifying patient clusters

  • how does this compare to Elizabeth Bromley etc (BPRS-PANSS)

Start with... Symptoms? Verbal Stuff? Non-verbal

  • Negative symptoms - 3 groups
    • 7 none (=7)
    • 7 mild (<= 10)
    • 6 moderate (> 10)
  • Positive and negative (how do the above groups spread across positive)
    • trend lower negative -- lower positive
    • NO correlation but small sample size (20)
    • Significant difference on pos symptoms between none and moderate neg symptoms groups
  • Neg - speaking
    • Speak less if more negative (trend)
    • Addressed more if more negative symptoms (significant)
    • Unaddressed more if mild symptoms than if none or moderate

- How is this different in controls?

- Use MRC study to see if these are general (non conext specific, 3-way etc) facts

  • Neg - non-verbal
    • More neg symptoms greater reduction in nodding when addressee (significant)
    • speaker gesture not significant but trend higher with more neg symptoms
      • nb more speaker gesture was related to less rapport
      • is more gesture because they are struggling etc
  • Neg - head-movement coordination
    • possibly less with moderate symptoms but too few to say
    • hard to interpret
  • Neg - rapport
    • More negative symptoms = less rapport
  • Neg - non-verbal sensitivity (PONS)
    • (did controls do differently?)
    • No difference
  • Neg - social cue recognition
    • Not really anything...?
  • Neg - social functioning (Social Network, MANSA, SIX)
    • Social network higher for mild symptoms than none or moderate (not significant)
    • is there a difference between mild+none v moderate?
    • Is age a confounding variable?
    • No diff in MANSA/SIX, but patients worse than controls

- Check power - Neg symptoms seem to be associated with the behavioural stuff

  • Neg - partners
    • partners nod less as recipient when patient has more neg symptoms
    • partners gesture more as speaker when patient has more neg symptoms
    • possibly aligning with patient

-- Case study group 17

13/06/2012 - Chris

Topics and word lists

Words that came out as most predictive for different things - comparing PEQ versus adherence

Looking at Adherence versus PEQ (total) Themes:

Adherence (e.g.) PEQ total (e.g.)
Illness related (objective) schizophrenic, symptoms Emotional/social (subjective) troubles, experiencing
Treatment regimen monthly, stable Physical symptoms overweight, bladder
Substance use lager, grass Domestic, leisure boy, sofa

For LDA topics, clear themes emerged; possibly relating to manually annotated topics; but is there a separation in the topics according to how doctors talk about e.g. symptoms compared to patients?

30/05/2012 - Mary

  • Background: Patients less likely to nod and gesture, or be speaker in interaction -- more likely to be the unaddressed

Q: Are they rejecting opportunities to take the turn or be addressee?

  • CA for Festschrift example
  • Segment of Group 23; A=patient.
    • Mainly between controls, with patient supplying minimal feedback; controls seem to have already decided between themselves who to throw off, then explicitly ask patient (select next speaker)
  • Patient: Gives feedback before being asked - doesn't take turns
  • Any substansive contributions?
  • Do patients prepare others for the fact they are going to start speaking (e.g. gaze, audible in breath etc)
  • Patient turns away when directly addresses her ("what about you?")
  • Male control also slightly awkward (but within range of 'normality')
    • Is male control not looking at to leave floor open so either can take next turn (i.e. not selecting next speaker and hoping patient may self-select)
    • Other control also does it in 28 (thinking out loud)
  • Patient seems disengaged despite apparently following with appropriate backchannels and non-verbal feedback before
    • "what, the" after long pause
    • Not that she's not following, but put on the spot
    • Lots of non-verbal attentiveness, and verbal agreement with their assessments, but not asked for original contribution off the bat
    • Rigidity in terms of how they make their contributions (less flexible in terms of moving from their point)?
  • Laughter; controls laughing patient isn't (is it nervous laughter or is it at her - or does she interpret it as at her even though it isn't)
    • Do controls laugh more all together or less (less nervousness?)
    • Who shares in laughter when it happens (here the patient doesn't twice in a very short segment)
  • Nice to see if same applies for those who also were in consultations
  • Also profile of patients; what symptom items etc
  • Time codes of start and end

23/5/2012 - Julian

Self-repair classification

  • Not exhaustive - but also want to narrow it down
  • Relationship between form and dialogue function (and correlations in terms of self-repair)
  • Transition space versus in separate utterance?
  • S-units versus U-units?
  • Pragmatic stuff of disfluency? Formal model (Ginzburg 2012) - forward-looking versus backwards-looking disfluencies (expansions or repair?)
  • Repairs marked in switchboard but not expansions
  • What is and what isn't repair?

Trying to get a semantics and pragmatics of self-repair - update type thing (what happens before and after)

  • Form--function
  • Need a strict definition of repair
  • Conflating disfluency and repair?
  • Things that haven't been done in terms of semantic processing and self-repair

Research questions: about formal model of self-repair and generation -- what should the semantic form be for input into generational module?

  • When does it get input to the generator?
  • Translation into surface syntax etc (different types of errors at different stages in generation)
  • Why and when do you need to repeat whole thing etc

Advantage of this corpus?

  • Disfluency marked - repeats
  • Too many categories?
  • Hierarchy of types?
  • Start by listing what they've done - how elaborate repairs are by dialogue type, what types of repairs occur in which dialogue acts etc
  • Incremental approach
  • Not face-to-face (justifiable in terms of dialogue models etc)

9/5/2012 - Mary

Balloon task non-verbal stuff

  • mm per frame female patients slower than male patients (m = 0.26, f = 0.21)
  • more patients movement (when speaking), less rapport controls report towards them
  • gesture / movement? - gesture defined as 2 sds over the mean of fastest hand movement - i.e. not observational
  • need to code for whether these are gestures or not (probably not?) but then what is it?
  • what else could it be if not 'gesture' (grooming behaviours, indicators of anxiety, etc)
  • if you are trying to engage the listener more you use more gesture?
  • more negative symptoms = more index of gesture -- try dichotomising the data between 'no negative symptoms' versus 'any' as small range etc
  • if patients are struglling to talk maybe rely on gesture more (as with L2 English speakers)

Looking at a dialogue

  • just hand markers
  • does her 'grooming' behaviour (fidgeting etc) count as 'gesture'
  • possible to look at combination of gaze and hand movement while speaking (she talks to her hands not the other people, but gaze des not include rotation of the head)
  • any way to index repetitive movements? stereotyped movements and not spontaneous
  • shoulder movements might be associated with these?

27/4/2012 - Sam

Music tuition - ethnography from Greenwich

Non-verbal interaction in music tuition in copresent situations as a basies for what goes wrong in video mediated tuitions

Focusing on a clip to highlight the importance of the score as a shared resource

21/3/2012 - Laura

Looking at a section of the transcript for CP261 in detail.

14/3/2012 - Chris

We went through some BNC transcripts looking at boundary cases (Schegloff) of P2 repairs.

Agreed that protocol is ambiguous/badly worded in some places - discussion of adding a separate category for the expansion type ones.

Points to note

  • Direct repetition is not proposed repetition
  • 'Initiator' shifts through protocol questions
  • Keep in mind that it is about intelligibility (though still questions of how expansion ones aid intelligibility by e.g. narrowing reference, these are not strict repairs of what was said in the case of the protocol - or at least, not necessarily so)

Everyone found the same cases hard to analyse, suggesting that there really may be some psychological reality in not treating all these as one category(?)

7/3/2012 - Rose

The CA technique of re-visiting the same data but with a different aim important. In this instance with more of an epistemics approach (in the sense of Heritage, 2005).

In this particular consultation, the doctor appeals to authority in several ways. Firstly, there is an appeal to insitutional practice/interactional ritual with "I just need t- to ask you some eh questions.. " (line 1).

There is coercive interrogator-like questioning, with reformulations/repeats of either alternative questions "do you really believe that or is it just an idea that comes into your head?" (21) or closed questions "it can't be possible can it?" (40), with a sense of increasing tendency to stronger refutation of the patient's experiences as being real.

Interestingly, one line of questioning finishes once the patient uses "I feel" (50).

Several cases of interruption by the doctor/increased volume (28, 40).

Interesting use of gesture with 'I feel' (50?), indicating something internal to her own experience.

Authority/institution based studies in Heritage and Rampton (2006)- the latter a sociolinguistic approach in schools.

29/2/2012 - Julian

Interested in building an incremental model of speech production.

Trying to build a model that can be implemented in a dialogue system

Categorising self-repairs to inform how speakers build up semantic representation, and cause.

Schegloff paper; function of repair - has to cover at least these examples (but interesting examples not real ones).

Self-repair in the switchboard corpus - majority = verbatim repeats? Is this because of nature of corpus.

Transcription issues...

Switchboard is telephone and people who don't know each other.

Articulation/repeat - different or not? Formulation

P3 repairs different to P1 repairs? How much of repaired bit gets recycled and what is similarity? {} fillers and discourse markers

Not all repair marked up... repeats and clear replacements - self-repair only and transition space/restarts might not be marked. Abandonment and restart.

Overlap not marked. No access to audio.

Classification versus function?

Some functions might be harder to judge (emphasis?) than others (vague-specific). Would some functions be dependent on the context of the repair? Interdependence of repair functions. Replacement/expansion not strict dichotomy etc.

Descriptive classification would be easier and more defensible. Mapping of form to function.

Word search ones - always marked by filled pauses? Reordering elements. Deletions, insertions (from CA). Post frmae/pre-frame name repairable (`what's it called').

Cross-dependencies of multiple repairs in single string; what repairs what (Schegloff e.g. 59)

15/2/2012 - Mary

Mary showed us one of her videos with three people (two controls and one patient) discussing the balloon task.

Possible things to pursue with regards to the linguistic bits (there are 40 untranscribed dialogues; 20 controls/20 with a patient) in no particular order were:

  • Grounding - patients do less nodding; do they also use fewer explicit spoken grounding cues (mmm, yeah, uh-huh)
  • Turn-taking - the patient's talk seemed to be frequently in overlap with one of the other participants. Do patients have issues with predicting when they may normally take the floor (resulting in more overlap or more gaps before they speak if the other participants are leaving them more time to take a turn)
  • Negotiating meaning - perseverance bias(?) - do patients engage less in accommodating their beliefs to others and stick with what they believe throughout the dialogues even where it is at odds with the other's beliefs. This may manifest itself in, for example, less lexical alignment from the patient.
  • Discourse markers - can't remember what the point of this was... Sorry

Links to papers I promised: Traum and Allen, 1992, A Speech Acts Approach to Grounding in Conversation

Clark and Schaefer, 1989, Contributing to discourse

8/2/2012 - Rose

We looked at a discussion between Dr/Patient and Husband

Questions that arose included

  • Party-formation in the interation - patient + carer? doctor + carer?
  • Differences in outcomes with added support - does this extend to additional person in interaction (any difference in types of support from extra person)
  • Doctors avoiding disagreement, so try to steer talk towards medication etc, but patients really want explanations for symptoms.

25/2/2012 - Laura

We looked at 3 examples of patient NTRIs to identify features of the contexts in which they occurr

Issues to consider/pursue:

- Lack of dr receipt/grounding/topic change- One possible environment for NTRIs is when doctors do not receipt a patient's turn and proceed with an unrelated line of questioning. The psychiatrist turn is therefore topically disjunct- the patient NTRI that follows is perhaps a result of a lack of grounding.

- There appears to be a negative relationship between MPCC scores (which examines aspects of doctor receipt) and frequency of NTRIS supporting the above (but no relationship between MPCC and adherence)

- May be useful to consider sequence organisation i.e. lack of post expansion/missing parts of sequences