top of page
Writer's picturePete Harvey

What do we remember when we remember a novel?

Updated: 3 days ago

In this article, I’ve attempted to summarise my PhD thesis - Recalling Atwood: Text World Theory and Memories of Narrative Fiction - for people outside my own academic field. I’ve also tried to answer the most important question of all: why does research like this matter?


Language Matters

What is the best way to write a succinct summary of a 90,000 word thesis? This article is written in a very different style to an academic publication. Obviously, lots of detail from my thesis is simply missing.  More significantly, the ideas and arguments that I have included are presented very differently. For example, I’ve tried to avoid using lots of technical vocabulary. When I have introduced technical terms I’ve tried to explain them succinctly, but this means I’ve inevitably lost some of precision with which these concepts are used in academic contexts. I hope I’ve done this without misrepresenting them too badly. However, this does raise an important question: are the ideas that I’ve presented here the same ideas that I presented in my original thesis? Or does my shift in style inevitably result in some fundamental change in meaning?

 

That question is relevant whenever anyone attempts to summarise academic writing for a non-specialist audience. However, it’s also a pretty good introduction to the topic of my thesis.


Why are narrative summaries interesting?

My PhD falls within the discipline of cognitive stylistics. This means that I’m interested in what happens in the mind when we speak and read – and in what specific language choices can tell us about the mental processes we use when we communicate.

 

For this reason, I believe that readers’ summaries of stories can be an important clue to mental processes that take place when we read and remember fiction. Summaries are often shorter than the texts that prompted them, and are usually presented in very different language, yet most people would probably say instinctively that a story and a story summary are fundamentally ‘about’ the same thing. But if they are, what is that ‘thing’ that they are about, and how is this distinct from the language of the original text?

 

This question is relevant to communication of all kinds, but it’s particularly relevant to how we understand fiction. When historians talk about their work, they will discuss entities such as Margaret Thatcher or the Houses or Parliament that existed in the real world. It’s usually possible, at least in principle, to decide whether or not the claims they make are accurate. However, when readers talk about George Orwell’s Nineteen Eighty-Four, they talk about entities such as Winston Smith or the Ministry of Truth which have no counterpart in the real world. Instead, readers imagine these entities in response to the language of Orwell’s text.  If a reader writes a summary of Nineteen Eighty-Four, and they describe the world of Orwell’s novel in a different way to the way that Orwell describes it, what are they really talking about – and can we decide whether the claims they are making are true?


What happens in our minds when we imagine fictional worlds?

Intuitively, we might think that we can make definitive claims about what happens in fictional worlds. Orwell’s Nineteen Eighty-Four opens with a memorable description of clocks striking thirteen. If someone later claims the novel opens with the clocks striking twelve, we can probably agree that they’re incorrect.


In practice, things are often not so simple.

 

This is because of a fundamental truth about language processing, which became evident when people first attempted to teach computers to speak. When humans understand language, they don’t simply rely on the words and grammatical constructions they hear or read. Instead, they combine linguistic information with what they already know about the world, including what they think they know about the speaker or writer and what they might be trying to communicate. So language processing is about ‘constructing’ meaning, rather than ‘decoding’ it.

 

Let’s show how that works with an example. Imagine I was telling you about a restaurant that had recently opened. I might say something like this:

 

(1) I went to the new Bistro in Sheffield last night. I didn’t leave a tip.

  Reading this, you can probably make a pretty good guess about whether or not I enjoyed my meal, even though I’ve made no statement to this effect. Indeed, I’ve not even told you explicitly that I had a meal at the Bistro, or that I did pay for what I ate while declining to add an optional service charge, though you probably inferred this. If you were to close your eyes, you might even be able to conjure a mental picture of a room full of tables and chairs, populated with waiters and managers, though I’ve mentioned none of these things. You may even find yourself able to make a fair guess about what the other people in this scene are thinking and feeling. In other words, comprehending my restaurant review involves you combining the information I have provided with what you already know about restaurants to create a rich mental representation of the scene I have described.

 

Why does this matter? Well, when people summarise novels, they often report on events they have inferred, as well as on events that are explicitly described in the text. For example, in one of the studies I ran for my PhD, I asked people to summarise a short passage about a woman who was informed of the death of her sister. Among 90 responses, four explicitly stated that the woman learned of her sister’s death by telephone, despite the original text making no reference to a telephone or telephone call.  This doesn’t mean that these participants misremembered the story, which never stated explicitly that the news wasn’t delivered by telephone. Instead, it suggests that they had drawn on their own knowledge of how bad news is delivered when creating their mental representation of the depicted scene.


In my study, participants added a telephone to their summary of a narrative, even though none was present in the source text.

 

This is just one example of how narrative summaries can reveal what’s happening in the mind when we read.

 

What can Text World Theory tell us about narratives?

One reason I’m interested in narrative summaries is because they can help us expand a theoretical framework known as Text World Theory.

 

Text World Theory builds on principles from cognitive psychology and cognitive linguistics, and offers a theory of the mental representations that people create in order to comprehend language. Unlike many other models of language comprehension, it makes explicit connections between specific linguistic cues and the different mental representations needed to understand an utterance. We can see how this works by looking at these two examples. I’ll call them micro-narratives for now. (Whether or not they really count as narratives is a question for another day!)

                                                                                                                                             

(2)    Pete ate an apple. He put the core in the bin.

(3)    Pete ate an apple. He might have put the core in the bin.

 

These two example narratives both prompt the reader to imagine a scenario (a fictional world) populated by a character named Pete and an apple. In the scenario prompted by example (2), this fictional world evolves as Pete eats the apple and then throws the core away. The scenario prompted by example (3), however, is different. Here, the presence of the modal verb ‘might’ prompts the reader to imagine two discrete worlds - one in which Pete eats the apple and one in which he throws the core away - and to understand that the state of affairs imagined in the second world may or may not occur in the first. In other words, example two requires the reader to construct two discrete mental representations and keep them separate.

 

Now let’s consider another example:

 

(4)    Pete ate an apple. He wanted to put the core in the bin.

 

Like the word ‘might’ in example (3), the verb ‘wanted’ primes the reader to conceptualise the state of affairs described in the following clause in a discrete world, and as a discrete mental representation. This time the relationship between the two representations is different. In (4), the scenario in which Pete puts the apple core in the bin is represents a state of affairs that Pete desires to bring about, rather than one which might have occurred. Again, understanding (4) requires the reader to keep the two discrete mental representations separate in their mind.

 

As Text World Theory outlines, understanding any narrative usually involves creating many such mental representations and keeping track of the relationships between them. This, in fact, this opens up a very different way of thinking about the purpose of narrative. Traditional theories tend to talk about narratives as semiotic representations of events unfolding in time. However, narratologist Marie Laure-Ryan has suggested that ‘tellable’ stories – those that people find most interesting – involve understanding not only what happened in a fictional world, but also how that world might have been different. We’re often told that conflict is essential to a good story – and we cannot understand conflict without being able to compare a world as it is with the world as the protagonist would like it to be.

 

Understanding even a short story involves creating and keeping track of many different mental representations (or ‘text-worlds’): including the ‘real’ world of a story[1], the conflicting worlds that different characters wish to bring about, and the accurate and mistaken beliefs that different characters might hold about the world at different points in time. This is a complex mental task, but it’s one we often do without conscious effort when we read. Text World Theory not only attempts to account for this cognitive effort, but also makes explicit links between what’s happening in a readers’ mind and the particular stylistic choices a speaker or writer makes when they tell stories.

 

What can Text World Theory tell us about how we remember narratives?

Lots of research exists on what happens when we try and remember something we’ve read. There is substantial agreement that even when recalling short passages, we usually fail to remember the exact words we read and instead remember the situations that those words describe. However, while true, that doesn’t tell us anything profound about how people remember narrative fiction, since even short stories usually run to several thousands of words.

 

However, Text World Theory gives us another way of thinking about stories because it allows us to ‘zoom out’ and look at fiction at a larger scale. If people don’t remember stories word-for-word, might they remember them world-for-world? And if not, can we start to think about why people sometimes remember only what ‘really’ happened in a story, and why they sometimes remember the different beliefs, desires and fears of different characters?

 

That, in short, was what I set out to explore.

 

So what did I learn?

I spent my PhD analysing reader summaries of novels – both summaries that already existed on book review sites likes Goodreads, and summaries that I gathered by asking volunteers to read and summarise short excerpts of novels under more controlled conditions.

 

One thing that is important to bear in mind when considering data like this is that the summaries I gathered did not give me direct access to a readers’ memories of a narrative. Writing a summary – whether for an online book review, or as part of an academic study – is a particular kind of task, and the conditions in which participants were writing would have shaped what they chose to include. I was very aware that any one of my participants would probably have been able to recall more details about the novel or experiment sample if I had asked them about it in a different way. For this reason, I was less interested in which details readers simply omitted from summaries. Instead, I focussed on what I could learn about the different configurations of text-worlds that readers recreated in their summaries – and in particular, on how these were often different from the texts that prompted them.

 

For example, I asked readers to summarise a passage from the opening of Margaret Atwood’s The Blind Assassin. I chose this extract because the narrator tells the reader very little about the world of the story, but a lot about the conflicting beliefs that characters have about the world. While some participants who summarised this passage were explicit about which characters believed what, others confidently reported particular characters’ beliefs as though they were fact. Summaries like this are not necessarily misreadings of the original text, but rather reveal clues about the (often unconscious) processes that readers use when they build mental representations of fictional worlds.

 

Another common strategy that many readers adopted when summarising novels was to attribute the beliefs or motivations of individual characters to wider groups. For example, Margaret Atwood’s Surfacing recounts the story of a woman’s journey home to find her missing father. In Atwood’s novel, the narrator travels in the company of a group of friends, who all have different motivations for making the trip. However, many readers produced summaries in which they described the group as a single agent with a single motive.

 

This was one example I noticed of a broader trend for people to assume that the beliefs or goals of individual characters were shared more widely. People would often summarise a passage describing the belief of a single character with phrases like ‘people believe…’, or ‘it is believed…’. So one strategy that readers appear to adopt when attempting to understand fictional words it to assume that the specific characters are representative of broader groups.

 

One more intriguing result emerged from the way that Text World Theory models ‘modality’ in language. The modal components of a sentence are those which encode the speaker’s attitude towards what they are describing. For example, in (2) above, the modal ‘might’ reflects the fact that the speaker is not certain about the truth value of the clause that follows.

 

Text World Theory distinguishes ‘boulomaic modal-worlds’ – mental representations of states of affairs a character desires to be true – from ‘deontic modal-worlds’ – mental representations of states of affairs which represent duties or obligations. This provides another possibility for differences between source texts and summaries. For example, in my Blind Assassin study, participants read the following line of dialogue when the protagonist and a policeman discuss what will happen after the crash that killed the protagonist’s sister:

 

(5) “I suppose you want someone to identify her,” I said.

 

Here, the verb ‘want’ means that the as-yet-unrealised state of affairs in which the protagonist identifies the body is conceptualised with a boulomaic modal-world. However, the majority of participants who referred this in their summaries did so by prompting a deontic rather than a boulomaic modal-world:

 

(6) She will have to go and identify the body

 

This sense that the narrator had an obligation to perform this unpleasant duty was one that was not conveyed in the language of the original text. However, the fact that so many participants made the same subtle change in meaning may be further evidence of them combining a textual prompt with their own existing knowledge as they read. This time, their prior knowledge appears to be shaping the relationships between the related text-world mental representations they need to build and maintain to understand the text.

 

In other words, detailed linguistic analysis of readers’ summaries of stories can provide clues to cognitive processes that take place when people read and later remember stories. These processes are not inevitable – it’s clear that different readers adopt different consolidation strategies on different occasions. However, exploring the range of consolidation strategies that are available can perhaps account for how different readers can arrive at different interpretations of the same text.

 

What next?

The bulk of my thesis focuses on exploring how Text World Theory can help us think about what happens when we remember fiction. However, I also touch on a couple more themes which I wish to expand upon in future research.

 

What exactly do we mean by ‘mental representation’?

Text World Theory provides a framework for exploring the different mental representations we create when reading and remembering fiction, but is less explicit about what form those mental representations take. In particular, it currently makes few predictions about whether readers’ representations of fictional worlds are built from mental pictures, from other kinds of mental images (such as mental evocations of sounds, smells or emotions), or from knowledge stored in the form of mental propositions. Exploring how specific linguistic cues lead to different kinds of mental representations will allow us to build a more comprehensive account of linguistic processing.

 

How does Text World Theory sit alongside other cognitive models of narrative comprehension?

Text World Theory is not the only framework which attempts to provide a cognitive account of what happens in the mind when we speak and read. In my thesis, I argue that it can be expanded by drawing on elements of another framework – the storyworlds model – and that together the two theories provide a more complete account of narrative comprehension.

 

So why does research like this matter?

Exploring how the mind processes language is a fascinating end in itself. However, I believe that in a world that is facing urgent problems, academics have a duty to think about how their research can contribute to finding solutions. Therefore I want to finish this summary by offering a couple of thoughts on how this research – and research like it – can be applied in the context of the climate emergency.

 

My research to date has focussed on how people understand novels. However, it’s an axiom of cognitive stylistics that we have not evolved separate cognitive processes for understanding fictional stories. Indeed, many people argue that constructing narratives is one of the fundamental ways we understand the world:

 

“A good story and a well-formed argument are different natural kinds. Both can be used as a means for convincing another. Yet what they convince us of is fundamentally different: arguments convince one of their truth, stories of their lifelikeness” (Bruner 1986: 11)

 

This has a direct impact on the challenges we face when talking the climate emergency. Much climate communication still relies on an ‘information deficit’ paradigm – a belief that informing people of the facts of anthropogenic climate change will be enough convince them of the urgency of the problem and prompt them to act. Yet this model is frequently criticised for being too simplistic. In this context, I suggest there is an urgent need to think more about the stories that people tell about the climate emergency and about the different roles that people might play in facing up to it.

 

For example,in the context of my own work on how people think about agency, I’m keen to explore more about how people think about who is responsible for processes which have positive or negative environmental benefits. Reading summaries of Annie Proulx’s Barkskins, I’m struck by how readers generally talk about activities such as starting a business and supporting a family as something individuals are responsible for, but generally talk about environmental destruction as something populations are responsible for. By contrast, ‘returning to nature’ is usually described as process that individuals complete on their own. Identifying patterns like this may provide clues about some of the unconscious assumptions that people draw on when thinking about climate destruction.


You can read my full thesis here:


And please do get in touch with questions.

 

For more thoughts about how studying language can help us explore how we conceptualise climate change and environmental destruction, here are some of the sources and projects that have shaped my thinking:

 

·       Stories we Live By – Aaran Stibbe

For Stibbe, ‘stories we live by’ are mental models we use to make sense of the world. They are often held unconsciously, but we can see evidence for them in the language people use. For example the story that ‘PURCHASE OF A PRODUCT IS A SHORTCUT TO WELLBEING is a story that shapes much advertising discourse. For Stibbe, challenging commonly held stories like this is key to helping people think differently about the climate emergency.

 

·       The Many Happy Returns Project

Many Happy Returns is an academic project focussed on reducing plastic waste. As part of a multi-disciplinary team, Emma Franklin, Joanna Gavins and Seth Mehl looked at the language people used to talk about plastic recycling, comparing the discourses of governments and businesses with the ways in which the public talk about plastic in discussion groups and on social media. Their analyses focussed on different ways of framing beneficial actions to influence more people, and on how metaphors such as THE CORPORATION IS A PARENT shape the way people think about consumption and recycling – is a brilliant example of how stylistics can have an impact in the world. They’ve produced a handy guide for businesses here.

 

·       Contemporary Fiction and Climate Uncertainty - Marco Caracciolo

With the growth of ‘cli-fi’ as a genre, many academics are asking what role novels might play in reframing the way we think about the climate emergency. Caracciolo makes a powerful argument that fiction – and in particular, novels that experiment with innovative narrative forms – can help us come to terms with the uncertainty of living in a world that is changing.


 

References

 

Bruner, Jerome. 1986. Actual Minds, Possible Worlds (Cambridge, MA: Harvard University Press)


Caracciolo, Marco. 2022. Contemporary Fiction and Climate Uncertainty: Narrating Unstable Futures (Abingdon: Bloomsbury Academic)

 

Elfenbein, Andrew. 2018. The Gist of Reading (Stanford, CA: Stanford University Press)


Gavins, Joanna. 2007. Text World Theory: An Introduction (Edinburgh: Edinburgh University Press)

 

Gibbons, Alison, and Sara Whiteley. 2018. Contemporary Stylistics: Language, Cognition, Interpretation (Edinburgh: Edinburgh University Press)

 

Herman, David (ed.). 2003. Narrative Theory and the Cognitive Sciences (Stanford, CA: University of Chicago Press)

 

———. 2009. Basic Elements of Narrative (Oxford: Wiley-Blackwell)

 

Ryan, Marie-Laure. 1991. Possible Worlds, Artificial Intelligence, and Narrative Theory (Bloomington: Indiana University Press)

 

———. 2019. ‘From Possible Worlds to Storyworlds’, in Possible Worlds Theory and Contemporary Narratology, ed. by Alice Bell and Marie-Laure Ryan (Lincoln and London: University of Nebraska Press), pp. 62–87


Stibbe, Arran. 2015. Ecolinguistics: Language, Ecology and the Stories We Live By (Abingdon: Routledge)

 

Stockwell, Peter. 2002. Cognitive Poetics: An Introduction (London: Routledge)


[1] The phrase ‘the ‘real’ world of a story’ is one I’m happy dropping into a non-academic article like this, but wouldn’t use in academic writing.  Fictional texts don’t refer to a ‘real’ world, so this is a tricky concept in theoretical terms. However, in my thesis I argue that many people read fiction as if it referred to some external reality – and we need to account for this in cognitive theories of comprehension.

14 views0 comments

Recent Posts

See All

Comentarios


bottom of page