Joep Cornelissen and Mario Kafouros, Leeds University Business School, May 2006
description
Transcript of Joep Cornelissen and Mario Kafouros, Leeds University Business School, May 2006
Coding and analysing metaphors within linguistic, psychology and
discourse perspectives: The use of quantitative methods
Joep Cornelissen and Mario Kafouros, Leeds University Business School, May 2006
Presentation structure
Three studies (‘linguistic’, ‘psychology’ and ‘discourse’) within organisation theory
Analytical steps around (1) identification, (2) inference and (3) analysis (systematicity/conventionality/use) of metaphors across corpus of language
Choices for methods Discussion: the potential of quantitative
methods and measures
Context
‘Organisations’ cannot be directly represented or experienced as single objects or entities (Sandelands & Srivatsan, 1993; Weick, 1989) metaphorically represented as organism, machine, (open) system, container etc.
Theoretical debates about how metaphors work and shape theorising about organisations
Very little empirical research on how metaphors are developed, selected and retained, and impact and shape the field
Studies funded by the Economic and Social Research Council (ESRC) (RES-000-22-0791)
Study 1: ‘Linguistic’ study
“why are certain metaphors selected and retained over time?”
A cognitive linguistic analysis of conceptual metaphors in organisation theory 1989-2003
hypothesised that four psychological principles (the between-domains distance principle, the within-domains similarity principle, the concreteness principle and the relational principle) determine the aptness of a conceptual metaphor, and its subsequent adoption and continued use over time
Data from academic publications 1989-2003
Principle Definition Operationalization References Between-domains distance
Metaphors that relate concepts from distant semantic domains
The principle is satisfied when the source concept comes from a domain other than ‘organization’; and then further divided into moderate distance (domains for collective human activities) and high distance (any other domain) 0 = non-existent 1 = moderate 2 = high
Blasko & Connine, (1993); Fauconnier & Turner (1998, 2002); McGlone & Manfredi (2001); Tourangeau & Sternberg (1981, 1982); Trick & Katz (1986)
Within-domains similarity
Metaphors where the correspondence between the target and the source concepts is conceived as more exact
The principle is satisfied when the source and target concepts share multiple salient features 0 = non-existent 1 = moderate (1-2 features shared) 2 = high (3 or more features shared)
Tourangeau & Sternberg (1981, 1982), Fauconnier & Turner (1998)
Concreteness Metaphors where the source concept that is referred to the target is concrete
The principle is satisfied when the source concept refers to concrete and tangible objects and events, rather than intangible or abstract ones 0 = non-existent 1 = moderate (concrete but intangible concept) 2 = high (concrete and tangible concept)
Katz (1989, 1992)
Relational Metaphors where the correspondence between the target and the source concepts is relational rather than attributive
The principle is satisfied when descriptions of the source and target concepts share relational structure rather than attributes 0 = non-existent (attributes only) 1 = moderate (attributes and relations) 2 = high (relational structure only)
Gentner & Clement (1988); Gentner et al. (2001)
Adjectives ‘organisational’ and ‘corporate’ as target terms into the Topic (title, abstract or keywords) field tag
(1) metaphor focus identification, (2) metaphorical mapping and categorization, (3) metaphor analysis
Ad 1: two coders and use of dictionary Ad 2: deductive and inductive, measured inter-rater
agreement (K-statistic) on categorization and on coding of aptness principles
Ad 3: measured frequency (total counts) and mention (spread over years) of metaphors over time, correlated with aptness scores
The kappa-coefficient (Carletta 1996) measures pair-wise agreement among coders making category judgments, correcting for expected chance agreement. Good quality categorization of discourse phenomena normally yields a K of about .80 (Carletta 1996).
Correlations between both the sum scores of aptness of conceptual metaphors and their frequency over time, and the scores of conceptual metaphors on each of the four principles and their frequency over time
Correlations between both the sum scores of aptness of conceptual metaphors and the scores of conceptual metaphors on each of the four principles, and their simple mention over time
Figure 1: Percentage of Conceptual Metaphors over the Total of Word Combinations 1989-2003
30%
40%
50%
60%
70%
80%
90%
100%
1989 1991 1993 1995 1997 1999 2001 2003
Year
Per
cent
age
Organizational
Corporate
Table 8: The Relationship between Principles Satisfied by Metaphors and their Aptness
Within-domains
similarity Between-domains
distance Concreteness Relational All (aptness) 0.28** 0.37** 0.41** 0.61**
Entire dataset (‘corporate’ and ‘organizational’
metaphors)
Dominant 100 (aptness) 0.84** 0.76** 0.59** 0.49** All (aptness) 0.17** 0.30** 0.40** 0.64** Organizational
metaphors Dominant 50 (aptness) 0.80** 0.71** 0.67** 0.38** All (aptness) 0.45** 0.47** 0.44** 0.57**
Corporate metaphors Dominant 50 (aptness) 0.87** 0.79** 0.53** 0.54**
** Significant at the 0.01 level Note: the correlations performed only include conceptual metaphors that have been mentioned more than once in the period 1989-2003.
Table 9: The Relationship between Aptness and the Retention of Metaphors
Measure 1: Frequency
Measure 2: Mention
All (aptness) 0.25** 0.00
Entire dataset (‘corporate’ and ‘organizational’
metaphors)
Dominant 100 (aptness) 0.53** 0.21* All (aptness) 0.25** 0.00 Organizational
metaphors Dominant 50 (aptness) 0.69** 0.00 All (aptness) 0.34** 0.02
Corporate metaphors Dominant 50 (aptness) 0.57** 0.24
** Significant at the 0.01 level * Significant at the 0.05 level Note: the correlations performed only include conceptual metaphors that have been mentioned more than once in the period 1989-2003.
Study 2: ‘Psychological’ study “What is the (psychological) impact of metaphors in
organisation theory?” The consequences or impact of a metaphor including
an ‘explicatory’ form of cognitive change whereby a metaphor facilitates learning or leads to conceptual clarification and a ‘generative’ type of cognitive change with a metaphor leading to conceptual advances and insights that were inconceivable before
Six central metaphors-in-use: organizational improvisation as jazz, organizational identity, organizational behavior as theatre, organizational learning, organization as chaos and organization as evolution.
250 participants (management and organization scholars) rated these metaphors on a number of scales
Participants were also asked to rate the effects of the metaphor in question; whether the metaphor had provided a language to communicate about a topic, facilitated learning/led to conceptual clarification, or led to conceptual advances.
P1: The higher the within-domains similarity of a metaphor, the higher the ‘generative’ and ‘explicatory’ consequences of a metaphor.
P2: The higher the between-domains distance of a metaphor, the higher the ‘generative’ consequences of a metaphor and the lower the ‘explicatory’ consequences.
P3: The higher the comprehensibility of a metaphor, the higher the ‘generative’ and ‘explicatory’ consequences of a metaphor.
Measures Used to Capture Constructs
CONSTRUCT MEASURE USED Within-domains similarity
To what extent (1) has it been easy to create correspondences between the 'organizational improvisation' and
'jazz' concepts? (2) have you experienced difficulty in seeing any resemblance between the subjects of
'organizational improvisation' and 'jazz'? (reverse coding) (3) do you feel that the ‘jazz’ concept captures important features of ‘organizational
improvisation’? (7-point scale ranging from ‘to a limited extent’ to ‘to a great extent’)
Between-domains distance
To what extent (1) do you feel that the entire domains ('organization theory' and 'performative arts') from where
the 'organizational improvisation' and 'jazz' concepts are sourced are rather alike? (reverse coding)
(2) do you consider these two domains ('organization theory' and 'performative arts') as rather distant in themselves?
(3) do you feel that these two domains ('organization theory' and 'performative arts') trigger very different associations?
(7-point scale ranging from ‘to a limited extent’ to ‘to a great extent’) Conceptual clarification
To what extent has this metaphor (1) organized existing knowledge on 'organizational improvisation'? (2) (re)emphasized features of 'organizational improvisation' that were already known? (3) led to conceptual clarification? (4) provided a useful new framework through which the act of 'organizational improvisation' can
be understood? (7-point scale ranging from ‘to a limited extent’ to ‘to a great extent’)
Conceptual advance
To what extent has this metaphor (1) led to inferences and testable hypotheses about the subject of 'organizational
improvisation'? (2) shown new features of 'organizational improvisation ' that were previously unknown? (3) offered new insights of 'organizational improvisation' that were inconceivable before? (7-point scale ranging from ‘to a limited extent’ to ‘to a great extent’)
Comprehension To what extent has this metaphor (1) made plain sense? (2) been easy to comprehend? (7-point scale ranging from ‘to a limited extent’ to ‘to a great extent’)
Descriptive Statistics, Reliability Coefficients and Correlations (N=439)
Variable Mean s.d. Reliabilitya 1 2 3 4 5
1. Within-domains similarity 4.7 1.9 0.71 - -0.34** 0.61** 0.46** 0.54**
2. Between Domains Distance 4.3 1.8 0.80 - -0.34** -0.14** -0.28**
3. Conceptual Clarification 4.7 1.6 0.77 - 0.64** 0.57**
4. Conceptual Advance 4.2 1.8 0.75 - 0.31**
5. Comprehension 5.1 1.6 0.68 -
a Cronbach-alpha internal consistency reliability coefficients * p.05 ** p0.01
Correlations for performative arts (jazz (N=31) and theatre (N=43)) and organizing
1 2 3 4 5
Theatre Jazz Theatre Jazz Theatre Jazz Theatre Jazz Theatre Jazz
1. Within-domains similarity - - -0.35* -0.35 0.68** 0.58** 0.54** 0.57** 0.37* 0.59**
2. Between-Domains Distance
- - -0.35* -0.57 -0.16 -0.21 -0.24 -0.25
3. Conceptual Clarification - - 0.77** 0.72** 0.39** 0.59**
4. Conceptual Advance - - 0.39** 0.36*
5. Comprehension - -
* p0.05 ** p0.01
Correlations for human psychology (identity (N=89) and learning (N=146)) and organization
1 2 3 4 5
Identity Learning Identity Learning Identity Learning Identity Learning Identity Learning
1. Within Domains Similarity - - -0.40** -0.38** 0.53** 0.63** 0.42** 0.53** 0.60** 0.58**
2. Between Domains Distance
- - -0.49** -0.38** -0.23* -0.19* -0.32** -0.29**
3. Conceptual Clarification - - 0.58** 0.65** 0.62** 0.64**
4. Conceptual Advance - - 0.39** 0.41**
5. Comprehension - -
* p0.05 ** p0.01
Correlations for biophysics (chaos (N=46) and evolution (N=84)) and organization
1 2 3 4 5
Chaos Evolution Chaos Evolution Chaos Evolution Chaos Evolution Chaos Evolution
1. Within Domains Similarity - - -0.29 0.31** 0.61** 0.69** 0.36* 0.52** 0.41** 0.57**
2. Between Domains Distance
- - -0.27 -0.31** -0.17 -0.12 -0.18 -0.29**
3. Conceptual Clarification - - 0.51** 0.71** 0.47** 0.58**
4. Conceptual Advance - - -0.01 0.36**
5. Comprehension - -
* p0.05 ** p0.01
Study 3: ‘Discourse’ study
Basic question in organisation theory – how people give meaning to events, work, environment within and outside of the organization
Massive literature on managerial and organisational cognition: schemata, cognitive maps, learning, scripts, mental models etc.
Most constructs strictly ‘cognitivist’ and ‘reproductive’ - emphasizing how the contents of cognition reflect, distort or otherwise mirror the world (e.g. computational metaphor)
Sensemaking as ‘productive’ account of meaning-making: how acts of cognition impose not only structure but also direction on experience (both in retrospective and prospective sense)
Sensemaking accounts, constituted through metaphors, link language (a discursive account of an organization as, for example, a machine), thinking (thoughts and ideas about what it means to see an organization as a machine) and action (acting as if an organization was a machine)
Need for systematic identification and analysis of metaphors in sensemaking/discourse processes
Discourse approach: (1) metaphor focus identification, (2) metaphorical mapping, and (3) metaphor analysis
Research context: corporate communications professionals within 6 organisations in the UK, asked to discuss their own experiences within a ‘critical incident’
First step: Interview data transcribed and analysed manually
Second step: use of a second coder for the categorisation (metaphor mapping)
Third step: cluster analysis (metaphor analysis)
Preliminary results
BOC: failed takeover by Airliquide 6 years ago “it was a time to re-engage the business (INTER-PERSONAL RELATIONSHIPS); we pointed out (VISIBILITY/VISUAL REPRESENTATION) to people that we were in control again, we had a new CEO who came (MOVEMENT) with a new message of growth and control, we organised brain-storm meetings to foster a can-do culture and to pre-empt employees bombarding (PHYSICAL AGGRESSION) corporate managers with questions”
BNFL: accident within the Thorpe plant “we needed to bring the message across (TRANSFER (OF OBJECTS) to local community, media and council as quickly as possible, we announced (DELIVER NEWS AS OFFICIAL MESSENGER) this first to the local media, when we talked to the local community we translated (LANGUAGE) it for them and explained how we saw the situation, we also told them that we weren’t yet in the possession of all the facts (POSSESSION AND TRANSFER OF OBJECTS), we consciously did not release information (PHYSICAL CONTROL OF OBJECTS) to the national press, in part so as to pre-empt negative news coverage”
Metaphor mapping: both deductive and inductive, K-statistic to measure inter-rater agreement on categorization of metaphorical foci into categories (source domains)
Metaphor analysis: measure frequency of particular metaphorical expressions and the frequency of source domains across the entire dataset, and in relation to samples related to a particular critical incident (as the sensemaking context), a speaker, and the organization involved.
Metaphor analysis: measure distribution of metaphors by sentence and consecutive ‘communications’ or ‘utterances’, as the unit of talk between the interviewer and the interviewee (Pollio & Barlow, 1975) (cumulative) frequencies, distribution
Discussion and implications
Quantitative methods at analytical points, and dependent on theoretical assumptions and research questions
Metaphor identification: use of quantitative measures not an issue in psychological research, but reliability and categorization measures (K-statistic, Perreault and Leigh statistic) important to linguistic and discourse research
Metaphor mapping and categorization: again kappa-statistic helpful for discourse and linguistic research
Metaphor analysis: measures dependent on research questions and dependent-independent variables (aptness, time) frequencies, correlations, distribution measures (poisson, cumulative frequencies, estimate functions)