Coherence is best described as the ‘impression a text leaves of being unified in some way’ (Charteris-Black, 2014, p. 55). When we encounter a text, we are left with the impression of a whole: it ‘hangs together’ (Halliday & Hasan, 1985, p. 48). Widdowson (2004, p. 63) describes the cohesion-coherence approach, introduced by Halliday and Hasan (1976, 1985), and later developed by Halliday (2004), as the ‘standard model’. In this paper I challenge the assumptions made by the ‘standard model’. The primary assumption I contest is described by Lukeš (2010, p. 183), who observes that meaning is thought of as something that ‘simply happens’. I adopt an interdisciplinary perspective based on the Neural Theory of Language (NTL) developed at the Institute of Cognitive and Brain Sciences at the University of California at Berkeley (Bergen, 2012, 2015; J. A. Feldman, 2006; J. A. Feldman & Narayanan, 2004a). NTL assumes that we ‘understand narratives by subconsciously imaging (or simulating) the situation being described’ (J. A. Feldman & Narayanan, 2004b, p. 389). Much like Lamb’s (1999, 2015) proposal for a neurobiologically plausible model of language, NTL maintains a commitment to the neuron doctrine’s extension to connectionist architectures (Freeman, 1999; McClelland, 1986; Yuste, 2015). I contribute to the NTL programme by integrating approaches developed in Cognitive Semantics (Fauconnier, 2009; Fillmore & Baker, 2010; Goldberg, 2006; Langacker, 1987), and by accommodating recent findings in Cognitive Neuroscience (see Kemmerer, 2014). I propose a theoretical framework that positions the experience of coherence as a gestalt simulation. I argue that the impression of coherence arises as a result of neural binding (Büring, 2005; J. Feldman, 2013; Roskies, 1999; Schmidt, 2009).

1. SIMULATING MEANING Terry

2. 1. Discourse and coherence
2. Taking the brain seriously
3. NTL and neurodynamics
4. Embodied Construction Grammar (ECxG)
5. Some Results
6. Next Steps
Sign-Posting

3. 1. THE ASSUMPTION OF

4. Defining Discourse
Language-use in ‘real’ situational contexts (van Dijk, 2007) – and “the
unfolding activity of producing and consuming text and talk” (Hart, 2014,
p. 4)
(i) The material practice of producing
text and talk (and its effects);
(ii) The neurobiological mechanisms
that underpin the material practice
of producing text and talk;

5. What is Discourse Coherence?
• First expressed by Hendricks (1967, p. 22) as the problem of “coherence
or unity”;
• To Halliday and Hasan (1985, p. 48): the way a text “hangs together”;
• To Charteris-Black (2014, p. 55): “the impression a text leaves of being
unified in some way”;
• A “relation between conceptual structures” co-present in text and mind
(Chilton and Hart, 2018, p. 121)

6. Approaches to Discourse
Coherence
Dominant Approaches:
Cohesion-coherence (n=19,269)
(Halliday and Hasan, 1976, 1985);
Discourse comprehension (n=8409)
(van Dijk and Kintsch, 1983)
Text semantics (n=4481)
(de Beaugrande and Dressler, 1981)
Other Approaches:
• Reference and co-reference (Behre, 1961; Winburn, 1962;
Halliday, 1964; Leech, 1965);
• Pragmatic realisation theory (Widdowson, 1978; Hobbs,
1979);
• Thematic Progression Theory (Danes, 1974; Fries, 1983,
1995);
• Rhetorical Structure Theory (Mann and Thompson, 1987,
1988);
• Relevance Theory (Sperber and Wilson, 1986, 1995, 2005);
• Causal connectives (Matt and Sanders, 2001; Sanders,
2005; Verhagen, 2007)
• Semantic coherence (Chang et al, 2009; Lau et al, 2015);
• Narrative comprehension (Emmott, 1999; Sanford and
Emmott, 2012);
• Conceptual coherence (Alonso, 2016);
• More: Tannen, 1989; Givón, 1993; Bublitz et al, 1999;

7. 2. TAKING THE BRAIN

8. Taking the Brain Seriously
Cognitive Linguistics ought to take the psychological and biological realities of
language-use “more seriously” (Dabrowska, 2016)
“…an important aspect of the functioning of our minds is to make themselves as
transparent as possible, keeping us from realizing that we are dealing directly only
with them, our cognitive systems, and only indirectly, and through them, with
reality.” (Lamb, 1999, p. 12)
While meaning is a result of intersubjective co-construction, the “mechanisms of
meaning” are entirely biological (Freeman, 1999, p. 9)
Against fragmentation and toward a reintegrated “linguistic science” (Christiansen
and Chater, 2017)

9. A Neural Theory of Language?
How does the brain compute the mind? (ICBS/ICSI, UC
Berkeley)
• Iterative, evidence-based, ‘best-fit’ approach
• Views language as “an embodied neural system”
(Feldman, 2015).
• “…we understand language by simulating in our
minds what it would be like to experience the things
that the language describes” (Bergen, 2012).
• Uses Embodied Construction Grammar (ECG) to
specify “schematic idealizations that capture
recurrent patterns of sensorimotor experience”
(Bergen and Chang, 2005).

10. Mesoscopic Brain Dynamics
Neuron population behaviours underlie EEG
wave activity in the cortex (Liljenström, 2012)
Observable: phase-state synchronicity across
neuron populations
Observable: recurrent phase-state transitions
between neuron populations (cortical dynamics)
Chaos (‘noise’) as dynamic complexity: neural
“states of meaning” (Freeman, 1999, p. 17)
Language is not represented in the brain;
language is a ‘raw’ I/O percept (Freeman, 2004:
pp. 3-4)
MACROSCOPIC (Topographic)
Regional anatomy
(cortical/subcortical)
*MESOSCOPIC (Intermediate)
Neuron populations, neuronal
clusters
MICROSCOPIC (Cellular)
Neurons, synapses, dendrites

11. Embodied Simulation
• Language understanding is based on the
“internal recreation of previous, embodied
experiences” (Bergen, 2012)
• Meaning as “centrally involving the activation
of perceptual, motor, social, and affective
knowledge that characterizes the content on
the utterances” (Bergen, 2012)
• Motor representations are activated by verb
usage (Glenberg and Kashak, 2002; Gallese and
Lakoff, 2005; Glenberg and Gallese, 2011)
• Neural integration and the “binding problem”
(Feldman, 2012)

12. A Neural Theory of Discourse
Coherence?
(i) IF neuronal wave activity generates phase-states and phase state
transitions:
(ii) IF neuronal phase states evidence the enactment of simulations:
(a) THEN enacted simulations are grounded directly in sensory experience;
(b) THEN coherence might be best understood as connectivity between
dynamic sensorimotor simulations;
(=/~) Do ‘linguistic signals’ express connectivity between conceptual
schematics?

13. 4. DESCRIBING COHERENCE: A

14. Embodied Construction Grammar
• CxG: form-meaning pairs (usage based)
• Deep semantic specification (Semspec):
embodied schemas (incl. metaphor)
• Parameterize active simulations
• Lattice-based configuration
• Computational formalism: ECG2
Workbench (NLU)
• Stochastic Petri (neural) nets
• Linguistic utterance/s evoke
enactment/s of simulation/s

15. Data sample
(1a) Liverpool FC would like to wish a Happy
Diwali and Bandi Chlor Divas to all our fans
celebrating the festivities today
(1b) Stop shoving Islam down our throat
(1a) Prepositional paraphrase of a ditransitive construction
(Goldberg, 1992):
X - - > Y Z
(1b) Intransitive construction with agent omission:
[X] - - > Y - > Z

16. (1) Coherence between role-
types (i)
• Schematic roles are “referring
expressions” (Bergen and Chang,
2003);
• Roles constrain and parameterize a
construction’s possible argument
structure;
• Role-type ontologies specify base-
profile relations
- Solid: explicit relation
- Dash: implicit relation
Expressed more formally as a relational
network

17. (2) Coherence between metaphorical
complexes (i)
(1a) Greeting as transaction
(A) -- > [WISH] -- > n=(∞)
(1b) Greeting as consumption
[WISH] -- > n=(B)
[WISH] < -- n=(B)

18. (2) Coherence between metaphorical
complexes (ii)
Reproduced from Bergen and Chang (2003)
X-schematic event potential(s): phase-states and phase-state transitions?

19. (3) Coherence between conceptual
primitives
[WISH] as trajector
- Landmark (fans)
- Container (body)
X-schematic event potential of [WISH] activates SPG
motion-event schema
- LFC (Source)
- Transacted signal (Path)
- “throat(s)” (Goal)
SPG as embodied sensorimotor primitive
- Localised in pre-motor cortex PMVc(F4)
- Hippocampal PbV subroutine
Reproduced from Bergen and Chang (2003)

20. 5. NEXT STEPS

21. Beyond the Thesis
Thesis represents the development of an ‘operational’ formalism (Lamb,
1999)
• Next step (1): psycholinguistic experiments (grading, scaling)
• Next step (2): neuroimaging and dynamic mappings
• Next step (3): computational implementations (Winograd Schemas)
• Next step (4): extensions beyond English
• Next step (5): context-specific applied variants
Always open to collaboration!

22. Contact: t.mcdonough@lancaster.ac.uk |
www.erotao.wordpress.com
Special thanks to:
Chris Hart
Paul Chilton
Aina Casaponsa
- Lancaster University
Jerry Feldman
Vivek Ragurham
- UC Berkeley