Conflicting processes in writing

Conflicting processes in
writing
David Galbraith
[email protected]
Southampton Education School
Writing as problem solving
(Hayes, 1996; Bereiter & Scardamalia, 1987)
 The thinking behind the text
– Retrieval of content from long-term memory
– Manipulation in working memory
 Knowledge telling v knowledge transforming
– Adapting to external rhetorical constraints
– Managing cognitive load
 Problem solving all the way down?
– Text production as local planning
– Passive output process
Dual process model
(Galbraith, 2009; Galbraith & Baaijen, 2015)
 Knowledge-retrieval process
–
–
–
–
Retrieval of ideas from explicit memory store (hippocampus)
Manipulation of ideas in working memory to create rhetorically
appropriate global model
Dependent on spatial component of working memory
Leads to creation of single knowledge object in episodic memory
 Knowledge-constituting process
–
–
–
–
Synthesis of ideas within semantic memory (neo-cortex)
Dispositionally guided text production
Sequential process, not dependent on spatial component of working
memory
Leads to formulation of ideas corresponding to writer’s implicit
understanding of the topic
Knowledge-constituting process
 Writer’s disposition = fixed connections between features in a
high dimensional semantic space (internal constraints)
 Ideas created by constraint satisfaction within network (content
synthesis)
 Successive utterances produced by inhibitory feedback from
output to disposition (self-movement of thought)
Summary
 Knowledge-transforming model
(Bereiter and Scardamalia, 1987)
– Single process associated with both
discovery and text quality
 Dual process model
(Galbraith, 2008; Galbraith & Baaijen, 2014)
– Spontaneous text production guided by
implicit organisation of semantic memory
– Construction of explicit mental model to
satisfy rhetorical goals
5
Relationship of processes with text quality
and discovery (Baaijen & Galbraith, under review)
 78 undergraduates writing article for
university newspaper (30 mins)
 2 different planning conditions (5 mins)
– Outline planning
– Synthetic planning
 Rate understanding before and after writing
(cf Keil, 2003)
 Two judges rated text quality
 Keystrokes collected using Inputlog
(Leijten & van Waes, 2006)
6
Keystroke analysis
 Two components identified from measures
of pauses, bursts and revisions
 Sentence production (α = .79)
– Long pauses between sentences combined
with clean bursts of language
 Global linearity (α = .80)
– Sequential production of sentences
7
Principal components analysis
Rotated Factor Loadings
Variables
1 Sentence linearity index
Component 1
Component 2
Global Linearity
Sentence Production
.849
.200
2 Percentage of I-bursts
-.830
.087
3 Percentage of time spent on events
-.824
-.009
.773
.075
-.709
-.270
6 Percentage of linear transitions between words
.601
.363
7 Percentage of bursts terminated by revision at
.113
-.901
.362
.809
9 Percentage of >2second pauses between words
-.072
.741
10 Text modification index
-.399
-.675
.181
.636
4 Percentage of linear transitions between
sentences
5 Number of production cycles
the leading edge
8 Percentage words produced in P-bursts
11 Mean pause duration between sentences
Eigenvalues
% of variance
α
4.7
2.4
42.47
21.40
.80
.79
8
Two independent writing processes
associated with discovery
Spontaneous text production
(knowledge constituting?)
Greater revision of global structure of text
(explicit reflection?)
9
Spontaneous text production
10
Relationships with text quality
 No relationship for outline planning
– Quality dependent on advance planning?
– (But outlining reduces discovery)
 For synthetic planning, higher quality
associated with more controlled sentence
production combined with revision of
global structure
 Conflicts with discovery
11
Relationship between sentence
production and text quality
12
Relationship between text quality
and discovery
 Sentence production effect on discovery
unrelated to text quality
 Global linearity effect depends on type of
planning
 For outline planning, discovery is associated
with poorer text quality
 For synthetic planning, discovery is associated
with better text quality
– Conflict between quality and discovery is removed by
synthetic planning
13
Relationship between text quality
and discovery
14
Conclusions
 Two conflicting processes
– Spontaneous sentence production
– Global organisation
 Outline planning conflicts with discovery
 Synthetic planning removes this conflict
because it allows global structure to
emerge during writing
15
Role of external representation
 The extended mind hypothesis
– (Clark & Chalmers, 1998; c.f. Olson 1994)
 Reflection on external object
 Does this play a role during writing?
– Dual process model predicts that it should
affect global organisation but not
spontaneous text production
16
Removing the external representation
(Galbraith & Baaijen, in preparation)
 77 undergraduates asked to write essays for
30 minutes about one of two topics
 Variables
– Vision v. no vision
– Notes v. full text
– Topic (drugs or terrorism)
 Compared lists of ideas produced before and
after writing (new v. old ideas)
– ideas rated for importance (global v. local)
17
Effects on discovery
Removing visual feedback reduces discovery because it preserves
original old ideas
Full text enhances the production of local new ideas and this is not
affected by visual feedback
18
Individual differences in writing beliefs
(Baaijen, Galbraith & de Glopper , 2014; White & Bruning, 2005)
 Transactional beliefs (TA)
– Degree to which explicit or implicit organisation is
prioritised
 Explicit control (low) v Implicit understanding (high)


“Writing helps me understand better what I’m thinking about”
“My thoughts and ideas become more clear to me as I write and rewrite”
 High TA discover more than low TA and
produce better quality text
– Outlining improves quality for writers with low TA
but not high TA; but reduces discovery to zero
– High TA revision is directed towards development of
understanding; low TA revision is directed towards
maintaining original plan
–
Effects of writing beliefs on
quality and discovery
Text quality
Discovery
20
Conclusions
 Intrinsic conflict between explicit organising processes and the
implicit organisation of the writer’s understanding.
 Outline planning favours explicit organisation at the expense of
understanding.
 Revision drafting strategy involving synthetic planning followed
by revision could reconcile the two processes.
 Revision directed towards development of understanding.
 Do new writing environments (media) disrupt development of
coherent knowledge object?
 Writing to learn should focus on learning goals (e.g. Wäschle et al.
2015; Galbraith, 2015).
21