Professional Documents
Culture Documents
Fergal Treanor
fergal_treanorhotmail.com
1. Overview
2. Theoretical issues
3. Practical issues
4. A note on intentions
5. Conclusion
6. References
1. Overview
Van Dijk (2006; 2008; 2009) presents a sociocognitive perspective on discursive context. His
multidisciplinary approach draws on findings from cognitive science and social psychology. At
the heart of his theory is the idea that context should be understood as an ongoingly updated
participant construct of the relevant contextual features of text and talk. He criticises traditional
sociolinguistic accounts of the discourse-context relationship as unnecessarily superficial, and
argues that the dynamics of this relationship are most usefully analysed in terms of the subjective
cognitive representations – “context models” – which discourse participants have of salient
contextual knowledge. These involve episodic and semantic memories, personal and
sociocultural knowledge, and procedural competencies that make interaction possible.
2. Theoretical issues
Van Dijk’s writing, however, does have a distinctly scientific tenor, and an analyst working in
the sociocognitive paradigm would certainly observe the tenets of constructive empiricism (van
Fraassen, 1980; 2001): study empirical data, draw inferences (about the phenomenology of other
minds), and present theoretically justifiable postulates (about what can be found there). Van Dijk
rightly rejects the empiricist fallacy in its narrow sense: unobservability is not seen as a limiting
principle (2006: 161). The ‘reality’ of these postulates can be accounted for by empirical testing
(2006: 164; 2008: 95,99).
Context models and the K-device, it seems, are understood to be ontologically anterior to their
detection via inference. They are hypothetical objects, whose existence can be inferred from their
correlates, and then tested for by means of specially designed experiments. To access context,
van Dijk argues, we should
systematically study its ‘consequences’, that is, discourse variations, in different situations, as we
do more generally in the study of unobservable phenomena in any science. (van Dijk, 2008: 107)
… without detailed experimental (and other empirical) studies I can only speculate on the ways
context models are formed, activated, updated and applied in actual discourse processes … what
follows are merely general hypotheses (van Dijk, 2008: 99)
This calls to mind the scientific ‘discovery procedures’ used, for example, in the discovery of the
planet Neptune, whose existence was first postulated, based on observed anomalies in the orbit
of Uranus, and then tested for by direct (telescopic) observation. This scientific approach aligns
van Dijk with the critical rationalism of Popper (2002), as it seeks to test hypotheses using
systematic empirical observation. Van Dijk does not explicitly call for falsificatory testing, but
presumably he would accept revisions to his model based on adequate empirical
counterevidence.
It seems unusual, then, that van Dijk likens the ontological status of these unobservable
phenomena to that of sociocultural abstractions, such as politics, society, etc. (2006: 163). This
comparison is slightly misleading, as while the subjectivities of other minds are taken to be at
least phenomenologically given, the other constructs mentioned here rely for their ‘existence’ on
the ontological primacy of observation, i.e. they are entirely our own constructs, for which the
analyst – or community of analysts – can claim full responsibility.
The phenomenological reading situates van Dijk’s postulates alongside Levinson’s “interaction
engine” (Levinson, 2006), as theoretically plausible components of our natural cognitive
makeup. It is analogous to the generativist idea of linguistic competence as a
“biological endowment … the innate component of the mind/brain that yields knowledge of
language when presented with linguistic experience” (Chomsky, 1986: xxvi)
It can in fact be argued that the K-device is a nativist concept; models of learned sociocultural
knowledge are emphasized throughout van Dijk’s theoretical elaborations, but of course, the
ability to develop subjective knowledge can be seen as objectively present in every (non-
pathological) human mind. Levinson (2006: 91) makes a similar point; regardless of how
relevant this “cognitive machinery” may or may not be to a given research programme, we can
acknowledge that it is there. To this extent, the theory is essentialist and universalising.
An important difference, however, is that van Dijk makes no claims about biological reality. He
makes it clear (2008: 65) that context models and the K-device are psychological, not
neurological objects:
“ … our analysis of the internal organization of mental models is framed in terms of schemas and
their categories, and not in terms of network structures, links, and the strength of such links – a
representation that might be closer to the neurological basis of mental models, but about which I
have nothing to say here.”
“crucial missing link that relates discourse processing to communicative situations and social
structures”
is unsatisfying, as it does not address the next question: where do participant constructs come
from? What is the foundation on which they are built? If the relationship between social structure
and discourse structure is superficial, surely psychological phenomena are also superficial in that
they are simply a manifestation of the underlying neurological reality. As our ‘next trick’, we can
argue that neurons and synapses are simply a manifestation of the interactions of molecules, and
so on, recurringly applying the principle that there is ‘something missing’ from our analysis – the
next level – something deeper – something more real. Luhmann (2009: 260-261) reduces this
argument to absurdity when he asks whether he must understand the cardiovascular function of
his critics before understanding their work.
To remain fair, however, van Dijk does not propose to depart from lifeworld phenomena or
ultimately to describe discourse in terms of particle physics. His essentially psychological
approach aims for the fundamentals of the lifeworld, the ‘atoms’ of experience. If we assume
that he intends to ‘get off’ there, then the ad absurdum criticism appears less relevant.
I do not share Luhmann’s dogma of irreducibility (Bohnen, 1994), but his perspective provides a
possible point of entry for the development of emergentist ontologies, which hold that while the
existence of lifeworld phenomena is predicated on beating hearts and an ‘m-class’ planetary
environment, these things do not saliently constitute the higher-order construct we call
“meaning”. In other words, “that which makes no perceptible difference by its presence or
absence is no real part of the whole” (Aristotle, cited in Givón, 2005: 179). Levinson (2006)
also posits three orders of discourse analysis – cognitive, interactive and sociocultural – which
can be understood as irreducible, and analysed accordingly in suitable ways.
My reference to critical rationalism is therefore not meant as a term of abuse, though in some
quarters of critical discourse studies, it might be regarded as such; it simply needs to be made
clear that the cognitive approach to discourse studies, while highly plausible, and heuristically
valuable, risks being paradigmatically and methodologically incommensurable (in the sense of
Kuhn, 1996) with more interpretative approaches to research; van Dijk’s constructive empiricism
is at quite a remove from the hermeneutic circle, and leaves less ‘scientific’ forms of thought to
the discourse participants under analysis. The approach is new, and the methods as yet
underdeveloped. Irrespective of this, I can agree fully with van Dijk’s premises, and concur with
his assessment:
3. Practical issues
Levinson (2006: 89) points out that no linguist would argue for the absurd idea that language
does not require mentation; the question does arise, however, of what form of analysis is best
suited to what ontological level of discursive context. Van Dijk himself mentions that context
models may vary only minimally when the activity type is institutionally constrained.
Indeed, we may assume that the more situations are formal, normative and institutional, the more
the context models of participants will be similar and overlapping, whereas those in informal
situations may be much more idiosyncratic. (van Dijk, 2006: 172-173)
Furthermore, as all the texts we are looking at are the products of institutions, or, where they are
written by individuals, by representatives of institutions, we can presume that the content of the
texts carries the approval of the institution-as-actor (Saarinen, 2008). This entails a
heterarchically distributed conception of agency, which is more usefully analysed as the
relationship between institutional structure and discourse, even if this relationship is
acknowledged in theory to be mediated by the confluence of the context models of many
different participants at many different times. Here, Husserl’s social ontology of mind provides
an illuminating theoretical backdrop:
“ …society, which we experience in a common consciousness, may be reduced not only to the
intentional fields of the individual consciousness, but also by the means of an inter subjective
reduction, to that which unites these, namely the phenomenological unity of the social life.”
(Husserl, 1927: 5)
Where the relationship between social context and discourse is concerned, I do think careful
empirical procedures will bring us a good part of the way; well-documented correlations and
certain minimal assumptions about the impact of extralinguistic structures, e.g. institutional
governance architecture, on linguistic action have proved quite useful in my own work so far.
What Levinson calls “the long-term sedimentation of interaction patterns” (2006: 92) is not
necessarily a question of cognition. Blommaert’s concept of “Orthopraxy” (2005) is quite an
elegant shorthand here, but needs to be elaborated. Interdisciplinary borrowing can help in
defining context, depending on one’s area of research. In my own current project,
neoinstitutionalist approaches to ideas and policy promise to be quite useful.
4. A note on intentions
“ … the formation of the intention of the current action… may be a fast and largely automatized
process, but that does not mean that cognitively it does not take place. Indeed … breaks in the
fluency of talk are also to be interpreted as manifestations of ‘ongoing thought’ within and
between turns at talk. There is more direct evidence when speakers actually refer to such ongoing
thoughts in conversation, e.g., when they say things like “Oh, I thought you meant . . . .” The
notion of intention is relevant for a theory of context because as a speaker or recipient I need to
construct myself as intentionally engaging in a communicative act… . That many of the aspects of
communicative acts are ‘automatized’ and barely conscious only means that mental models are
partly processed in the background, as is also the case for context models.” (2008: 82)
This appears to imply that ‘it’s intention all the way down’; that the quality of being intentional
always applies to a communicative act as long as it was not a typo or a slip of the tongue. But
can we truly characterize intentions simply by awarding a one or a zero in this way?
The kind of metapragmatic phenomena van Dijk refers to here furthermore indicate that
interlocutors do not, in the normal flow of discourse production, have conscious control of their
choice of formulations. To pause and actively reflect is to move, mentally, to a different level of
awareness of one’s own intentions. I argue that in institutional contexts, the unnamed authors of
texts are situationally constrained in that they are not at liberty to engage in such metapragmatic
activity; at any rate, they cannot reproduce it in official documents.
5. Conclusion
This paper explicitly accepts the value of the sociocognitive perspective, and is far from being
‘anti-mentalist’. I do however call into question the value of the sociocognitive research
paradigm in tightly constrained institutional settings.
In other words: while I can share van Dijk’s ontological assumptions about psychological objects
in the lifeworld, I am not fully convinced that these assumptions need lead us to a radically new
methodology.
This view applies only to discourse analysis in settings with stable contextual factors; the
cognitive view could prove very useful in designing research programmes that are more strongly
focused on individual subjectivities, e.g. verbal interactions, experiments in social psychology.
It can be objected that my reluctance to adopt the sociocognitive approach is a license for
fogginess; instead of being guided by a strong scientific theory, I am simply embracing the
relativity of the postmodern world view, and thereby failing to make progress towards scientific
truth, or even towards increased empirical adequacy.
In answer I can only restate that interpretation is not ad hoc, but driven by data and theory.
Subjectivity does not preclude rigour. Even in the most scientific disciplines, subjectivity is
always present. By acknowledging this explicitly, analysts invite readers to participate in the
joint construction of intersubjectively sound interpretations, the best of which will harmonize in
productive ways with findings from adjacent disciplines.
6. References