Literary meaning as character conceptualization: Re-orienting the cognitive stylistic analysis of character discourse and Free Indirect Thought

1 Cognitive linguistics and the analysis of mind

Cognitive linguistics provides the best toolkit for mining the meaning of literary language for a deeper understanding of how characters’ minds work. It is a broad field of theories and frameworks which share a common objective: understanding language in terms of more general cognitive processes. Lakoff (1990: 40) states that research within cognitive linguistics is beholden to the ‘Cognitive Commitment’, which demands a characterization of language ‘that accords with what is known about the mind and brain from other disciplines’ (see also Evans 2012). The different theories and frameworks that make up the field explore how language relates to various aspects of cognition. For instance, Conceptual Metaphor Theory examines how the mind organizes knowledge and experience according to metaphorical relationships which underlie and motivate linguistic expressions (see Kövesces 2010). Cognitive Semantics examines the overlapping structures between language and other conceptual and perceptual systems, especially vision (Talmy 2000). Construction Grammar explains syntax and morphology in terms of knowledge that is cognitively organized into semantic patterns at various levels of abstraction (Hilpert 2014). And so on. For all cognitive linguistic theories, language is not a fundamental or separate module of human cognition; it is inextricably dependent on and related to other cognitive processes and conceptual phenomena in the mind.

This article will draw most heavily on the framework known as Cognitive Grammar, hereafter CG, as it is laid out in Langacker (1997, 2008). CG appropriates many ideas from cognitive psychology and other cognitive linguistic theories in order to develop a unique understanding of grammar – in the broadest sense of the term – in terms of the cognitive processes wherein it obtains its meaning. Meaning, in CG, is a property not just of lexical items but of also of abstract grammatical categories and morphosyntactic constructions. Langacker equates meaning with ‘conceptualization’, which in turn he defines broadly as ‘mental activity’ (2008: 119). The semantics of all aspects of language, from morphemes to parts of speech to larger syntactic units, are characterizable in terms of the mental activity associated with them. CG can be distinguished from Construction Grammar in that it is not as concerned with identifying different syntagmatic constructions and their semantics. Rather, it is more focused on the cognitive processes that motivate linguistic forms and underlie the ‘phenomenological experience’ of linguistic meaning (1997: 51). It is this analysis of language in terms of underlying mental activities that makes the framework so appealing for the exploration of minds in fiction.

The mental activities that concern CG are operating at a level of cognition that is both prior to language itself – in the sense of not being based in or dependent on mental language – and often below the level of conscious awareness. The most important general type of mental activity that concerns CG is what Langacker calls ‘construal’ (2008: 4). Construal refers to our ability to conceive the same conceptual scene in different ways, largely in terms of how we direct our attention within it. It can concern, for example, what we focus our attention on, at what level of specificity, from what perspective, and the degree to which we attend to our own processes of conceiving. According to Langacker, construal is ‘roughly analogous to adjusting the focus on a pair of binoculars: it affects the viewing apparatus (the mind) but is not itself a target of observation’ (1997: 54). Thus, a speaker or hearer is ‘not usually conscious of […] imposing a particular construal’ on the content evoked by a linguistic expression (54), which leads to the conclusion that the grammar of an expression ‘is shaped as much by what we are not consciously aware of as by what we are’ (74).

CG – like other cognitive linguistic frameworks – provides an enticing tool for the analysis of discourse precisely because it can be used to unveil the non-linguistic and non-conscious mental activities that underlie language use. Lakoff’s ‘Cognitive Commitment’, as mentioned above, commits the discipline to characterizing language based on what we know about the mind from cognitive psychology and neuroscience. However, Evans (2012: 8) claims that the knowledge transfer can also go in the opposite direction, that language can serve as ‘a lens for studying aspects of the mind’. Langacker (2008: 4), similarly, insists that grammar ‘is not only an integral part of cognition but also a key to understanding it’. These ideas are clarified by Hart (2011: 173):

Cognitive Linguistics does not claim that it can ‘look’ directly into the brain [via linguistic analysis]. Rather, it argues that, since communication necessarily involves conceptualization, and language is based on the same system that we use in thought and action, linguistic structure affords indirect access to conceptual processes. In this sense, language is for Cognitive Linguists ‘a window to the mind’ (Fauconnier 1999).

Thus, just as we use the cognitive and brain sciences to inform our understanding of language, the analysis of language can inform our understanding of what is going on in the mind in terms of both phenomenological experience and cognitive processes beneath the level of awareness.

This prospect brings up an important question: if cognitive linguistics can be used to treat language as a lens into the mind, whose mind should we be looking into? At a theoretical level, cognitive linguistic frameworks like CG are mostly concerned with describing a linguistic system in relation to human cognitive processes in the abstract, as they pertain to the species in general. However, such frameworks can also be applied to the analysis of specific instances of language use, i.e. discourse, which is no longer concerned with the abstract human mind but with real minds that belong to the individuals involved. In the existing scholarship, different types of discourse analysis have applied cognitive linguistics in relation to different discourse participants. Tenbrink (2015), for instance, in her ‘Cognitive Discourse Analysis’ methodology, focuses on language producers. She conducts experiments that elicit highly controlled responses from subjects in order to determine how ‘the speakers’ linguistic choices will necessarily reflect their conceptualization of the scene in systematic ways’ (4). Hart (2011, 2014, 2015), meanwhile, focuses more on language receivers. In his application of cognitive linguistics to Critical Discourse Analysis, his primary concern is with identifying ‘the effects of ideological or perspectivized language use on hearers’ mental representations and evaluations of reality’ (2014: 9); although he also recognizes the validity of a speaker-oriented approach.

In the application of cognitive linguistics to the analysis of literary discourse in cognitive stylistics, most scholarship has similarly focused on analyzing meaning in terms of readers’ mental activity (e.g. Browse 2018; Culpepper 2001; Emmott 2001; Giovanelli 2013; Giovanelli & Harrison 2018; Harrison et al. 2014; Harrison 2017; Nuttall 2018, 2019a, 2019b; Stockwell 2009, 2014). This is in keeping with a dominant trend in broader stylistics towards reader-response criticism. According to Semino and Culpepper (2002):

[By] focusing on the relationship between linguistic choices and effects, stylistics has always been concerned with both texts and readers’ interpretations of texts. […] What is new about cognitive stylistics is the way in which linguistic analysis is systematically based on theories that relate linguistic choices to cognitive structures and processes. This provides more systematic and explicit accounts of the relationship between texts on the one hand and responses and interpretations on the other (2002: xi).

Stockwell (2009), gives the reader an even more central focus within the field he calls cognitive poetics, using cognitive linguistics and cognitive science to account for real readers’ experiences of texts. He argues that ‘the proper business of literary criticism is the description of readings [which] consist of the interaction of texts and humans’ (1). Writers, off course, are also real humans that interact with a text; and while some scholars acknowledge that writers’ minds can be the focus of cognitive linguistic analysis (e.g. Harrison et al. 2014), this is rarely the case in cognitive stylistics/poetics. Giovanelli (2019) is an exception, analyzing poetry in direct relation to its contexts of conception and authorial creativity. Meanwhile, Stockwell (2009) and Stockwell and Mahlberg (2015) treat authors’ minds primarily as reader interpretations.

However, literary fiction brings into existence another type of individual in relation to whose mind linguistic meaning can be analyzed: a fictional character. In what is often termed ‘mind style’ analysis, stylisticians determine how linguistic patterns create the ‘distinctive linguistic representation of a [character’s] individual mental self’ (Fowler 1977: 103) or ‘the impression of a worldview’ (76). Some cognitive stylisticians have, in certain parts of their mind style analyses, explored semantics in direct relation to a character’s mental activity (e.g. Giovanelli 2018; Nuttall 2014, 2018; Semino 2002; Semino and Swindlehurst 1996). However, such character-focused passages are almost always embedded within a more overarching concern with how the reader attributes or ‘models’ character minds (Stockwell and Mahlberg 2015: 134; see also Browse 2018; Clark 2009). Character minds are thereby usually treated as by-products of readers’ processing of texts. A much rarer type of mind style analysis in cognitive stylistics is one that determines linguistic meaning directly in relation to characters’ minds to the extent that real readers are backgrounded and all but ignored (e.g. Rundquist 2020). Even though character minds are obviously linguistic constructs that have no real existence outside of readers’ and authors’ imaginary experience, this last type of analysis treats fictional minds as if they have an underlying depth of mental experience and cognitive processing that is as relevant for linguistic meaning as that of the real-life language producers and receivers.

2 Grounding and character conceptualization

While it is this character-oriented approach that is advocated in the present article, all of the analytical approaches described above are justifiable if we examine the nature of semantics in Cognitive Grammar. Meaning, it turns out, can be analyzed in relation to textually constructed characters, as well as real life discourse participants.

For Langacker (2008), as discussed, meaning is conceptualization, i.e. mental activity, of which phenomenological experience is the tip of the iceberg, underlaid by unconscious cognitive processes. For any given linguistic expression, the individuals whose mental activities constitute meaning are known as the ‘conceptualizers’ (73). In Langacker’s explanation, conceptualizers are prototypically identified with the ‘speaker’ and ‘hearer’ of an expression, the extratextual discourse participants who exist in the real world and whose minds process the language and create meaning (73). Thus, linguistic meaning is prototypically understood to be constituted by the mental activity of the real people who cognitively process it. However, crucially, a conceptualizer is also a linguistic construct within the text itself; and this means that it can be dissociated from those extra-textual participants. Whenever the language of an expression identifies someone else as the conceptualizer, such as a fictional character, then conceptualization – the mental activity that constitutes linguistic meaning – can be attributed to them instead. This is something that Langacker discusses briefly in terms of the ‘fractal organization’ of conceptualization (484), but it can be explained more precisely with his concept of ‘grounding’ (259).

Grounding is one of the ways in which conceptualizers manifest linguistically. It occurs primarily through what Langacker calls ‘grounding elements’, a small set of closed-class features that implicitly link the discourse to the conceptualizers’ subjective contexts and online conceptions of reality (2008: 259). These include articles and demonstratives, which indicate the conceptualizers’ familiarity with and proximity to nominal referents, and verb tenses and modal auxiliaries, which indicate the relative time of occurrence and epistemic status of verbal processes relative to the conceptualizers’ subjective viewpoints. Grounding also occurs through a broader range of features that more explicitly mention certain aspects of the ground, such as the conceptualizers themselves, their subjective ideas, and times and places relative to their own contexts. These include deictic pronouns and adverbs (e.g. “I” “you”, “here”, “tomorrow”), modal adjuncts, and overtly subjective content like evaluation and expressive syntax. In prototypical discourse, all these grounding features provide referential links, with varying degrees of explicitness, between the language of an expression and the extratextual context of the speaker and hearer. They also identify the speaker and hearer as the conceptualizers whose mental activity constitutes the linguistic meaning. Thus, because of grounding, in prototypical discourse, conceptualizers are both textual constructs and extratextual participants. However, literary fiction is not prototypical discourse.

In fiction, the linguistic reference to conceptualizers through grounding features can be dissociated from the extratextual contexts of the real-life speakers and hearers. This is what happens when a character is quoted or when they serve as the narrator of a story: the language is grounded in their subjective context as if they were the extratextual speaker, even though they have never existed outside of the text. The character is therefore a non-prototypical conceptualizer: even though they do not conceptualize the discourse meaning (simply because they do not exist in real life to do so), the discourse reflects their mental activity as if they did. Therefore, for all practical and analytical purposes, the language can be understood to derive its meaning in terms of their underlying mental activity.

To be even more precise – and at the risk of overstating the obvious – within a character’s discourse, grounding features like ‘I’, ‘here’ and ‘now’, evaluative features like ‘stupid’ and expressive constructions like questions and exclamations, are all obviously interpreted to reflect the character’s subjectivity and conceptualization. This interpretation of meaning relative to the character’s mind includes other features that may not be explicitly subjective but are of interest to cognitive linguistics, e.g. semantic frames, constructions based in conceptual metaphors or conceptual blends, linguistic construals, etc. Even though characters obviously do not actually experience the underlying mental activities associated with which such features, the language obtains its meaning in relation to the fictional occurrence of such mental activities. Cognitive linguistic analysis therefore can treat a character’s discourse as a lens into that character’s mind, as was done in many earlier, pre-cognitive mind-style analyses (e.g. Fowler 1977; Halliday 1971).

A reader-response oriented cognitive stylistician can legitimately question the basis for analyzing linguistic meaning in terms of the underlying mental activity of a fictional character. Meaning, they might argue, is not a property of the text itself but of the minds of the real people who process and produce it; and a fictional character’s mind is only a textual construct, an epiphenomenon that depends entirely on a reader’s interpretation or an author’s conception. This is a valid concern. However, meaning is conceptualization; and conceptualization, as it is understood in Cognitive Grammar, is a two-way street. It pertains not just to the minds that conceive and process the language but to the minds whose mental activity that language purports to reflect or evoke. In fiction, these conceptualizers need not be equivalent. It is of course possible, if not likely, that the character’s underlying mental activity exists at some level in the reader’s and author’s minds as they process the discourse and ‘enact’ the character’s mind (Caracciolo 2012: 45). The author and reader are unlikely to be aware of things like grammatical construals, framing strategies or underlying metaphors, but they may nonetheless perceive them ‘subliminally’, or as an effect of ‘tone’ (Fowler 1977: 76). A character-focused analysis can therefore indirectly reveal a great deal about the reader’s experience or author’s intentions.

Analyzing a character’s discourse in terms of its representation of their own mental activity is simply a matter of analytical focus and preference. When an analyst focuses instead on a reader’s mind, they prioritize a certain degree of interest in how the human mind processes language while reading, including how it interprets characters’ minds. If meaning is analyzed relative to an author’s mind, the analyst is more concerned with the context, intentions, background knowledge, and other mental activities in the process of language production. When we analyze a character’s discourse in terms of that character’s own mind, on the other hand, we are prioritizing a deeper understanding of what their non-linguistic and non-conscious mental activity would be like if they were to produce that discourse as speakers in a fictional world. As we will see in the next section, however, the analysis of a character’s underlying mental activity is not limited to that character’s communicated discourse.

As a side-note, it should be acknowledged that when discourse is quoted in non-fictional contexts there is a similar type of dissociation between extratextual and intratextual conceptualizers. However, in that case the quoted speaker’s extratextual conceptualization is still presumed to have existed in real life, only in a different time and context; otherwise, it too is fictional. Non-fictional quoted discourse, unlike fictional characters’ discourse, is therefore analyzable in terms of two contexts of real-life extratextual conceptualization, it is just that one of them happens to be ‘displaced’ from the current discourse context (Chafe 1997: 359).

3 Thought quotation

Speakers and hearers and writers and readers, whether fictional or real, are not the only types of conceptualizers: it is also possible for language to evoke conceptualizers who are not communicating at all, even within the fictional world. This happens when narrative discourse gives direct access to the private conscious experiences of non-speakers through the quotation or representation of their mental content. Such discourse situations are inherently fictional: outside of fiction, one cannot directly (i.e. non-inferentially) observe the precise contents of another person’s mind if they do not communicate them. Thus, the presentation of fictional consciousness creates a unique context for cognitive linguistic analysis to illuminate a conceptualizer’s mental experience in ways that are potentially more direct and profound than the analysis of their communicated language.

One type of mental experience is that of verbalized thought, or ‘inner speech’ (Palmer 2002: 31); and while this is not the only mode of human thought, it is the only one that can be directly quoted in literature. This is done with the thought presentation categories known as Direct and Free Direct Thought (here conflated as F/DT). When used to quote inner speech, F/DT achieves the ‘pure mimesis’ of thought content (McHale 1978: 258), ‘imitating’ an aspect of a character’s private mental experience with maximal acuity (Plato 1994–2000). This technique is inherently fictional. While humans have a Theory of Mind which enables us to productively infer the contents of others’ thoughts in real life (see Herman 2011), this does not normally extend to the exact language of those thoughts; moreover, it is purely inferential. Any use of F/DT to quote someone else’s thoughts outside of fiction is bound to be pretense, or at best an approximation. Even attempts to communicate one’s own private inner speech aloud would negate the possibility that it either ‘private’ or ‘inner’. In fiction, the quotation of a character’s inner speech is also, of course, pretense; but because it occurs within a fictional context it can be intended and understood as an exact, non-inferential quotation of the character’s mental content.

F/DT is fully grounded in the character’s subjective context with the same types of grounding features that are found in quotations of their communicated discourse, e.g. deictic, evaluative and expressive features. These identify that character as a conceptualizer for the cognitive linguistic analysis of meaning in terms of their underlying mental activity, in the same way as communicative speakers. However, the nature of inner speech makes them a different sort of conceptualizer, one that neither communicates nor comprehends the language of others but merely experiences the self-directed language of thought. Due to this unique context, the cognitive linguistic analysis of inner speech is likely to illuminate character psychology in ways that are, if not more profound, then at least epistemologically distinct from the analysis of communicated discourse. Inner speech is less likely to be clouded by the rhetorical or manipulative strategies of interpersonal communication and therefore likely to be more directly reflective of underlying cognitive processes (see Nuttall 2018: 21, 85).

However, inner speech is also a fairly surface-level cognitive phenomenon. Human ideation is far more complex and varied than a simple stream of internal monologue (see Carruthers 2002; Damasio 1994; Palmer 2004; Pinker 2007). Thought has a wide range of phenomenological ingredients – what James (1890: 147) calls ‘mind stuff’ – which might include things like pre-verbal concepts, visual images, other perceptual memories and projections, image schemas, and bodily movements and sensations, in addition to fully formed language. Words are not the ‘medium we think in, but a tool we think with’ (O’Brien and Opie 2002: 694). In literary fiction, a character’s non-verbal thought content is obviously not available for purely mimetic quotation in the same way as inner speech; it is, however, accessible through techniques of mimetic representation. Mimetic representation makes non-verbal thought accessible to mimesis, allowing literature to achieve higher standards of ‘psychological realism’ and creating some of the most interesting contexts for analyzing characters’ minds.

The most straightforward and prototypical technique for achieving mimetic representation is what is known as Free Indirect Thought (FIT), although other techniques may also have this semantic potential in certain contexts. The next section will explain the grammar and representational semantics of FIT. Then the following section will use that understanding of FIT, together with that of characters’ quoted discourse discussed above, as a basis for subdividing the notion of ‘conceptualizer’ in CG into distinct types. This will validate the cognitive linguistic analysis of meaning in relation to the underlying mental activity of various individuals who are involved in discourse in different ways, including fictional characters whose minds are merely represented by the language. The final section will demonstrate a short analysis that applies cognitive linguistics directly to the mind of a non-speaking character that is represented with FIT and FDT.

4 Free Indirect Thought

Free Indirect Thought can be defined as narrative language that expresses a character’s subjective thought content alongside narratorial deictics for tense and person and which occurs unsubordinated to a narratorial reporting clause. In Cognitive Grammar terms, FIT is a case of dual grounding: it is grounded in the narrator’s context with tense and personal pronouns, while it is also grounded in the character’s more overtly subjective context with features like articles and demonstratives, modality, evaluation and expressive constructions. The following is an example of FIT in Ian McEwan’s novel Amsterdam (1998), where the protagonist is contemplating the success of a recent business decision. The features that are grounded in the character’s subjectivity are underlined while those with narratorial grounding are in bold:

It was goingwell, it was going well. But for this one little thing hewould be hugging himself, hewould be dancing on the desk. It wasrather like this morning when he lay in bed contemplating his success, denied full happiness by the single fact of Clive’s disapproval (117).

The character-oriented subjective features include the ‘deictic’ use of progressive aspect (Adamson 1995: 26), evaluation, repetition, deictic determiners, modal auxiliaries, intensifying adverbs and emphatic adjectives. These features indicate, with varying degrees of explicitness, the discourse’s grounding in the protagonist’s subjective context, in the same way that they would if he were the fictional speaker. At the same time, the past tense and third-person deictics also partially ground the discourse in the narrator’s context. Thus, FIT, like F/DT, has mimetic qualities: by grounding the discourse in the character’s subjective context, it has the effect of imitating his thought content rather than describing it from an external perspective. However, the narrator-oriented deictics make it clear that the language cannot the pure mimesis of thought quotation; it must originate at least in part from the diegetic level of narration. This leads Toolan (2001: 135) to describe the style as a sort of ‘mimetic diegesis’.

There is a long history of debate in stylistic scholarship about how FIT relates to the discourse and subjectivities of characters and narrators (see Rundquist 2017, for an overview). These range from notions of dual voices (Pascal 1974) and partial quotations (Ehrlich 1990) to sentences without speakers (Banfield 1982), each of which has some degree of validity. However, the most straightforward way of understanding FIT is as a mimetic representation of something ‘like’ what is going on in the character’s thought (Banfield 1982: 80). This is because the combination of character and narrator deictics creates an inherent ambiguity about the fictional world phenomenon being conveyed. In the context of Free Indirect Speech, this ambiguity simply casts into doubt the character’s precise utterance in the fictional world. For FIT, on the other hand, it opens up the possibility that at least some of the thought content represented is pre-verbal in nature. Thus, the dual grounding of FIT is a powerful semantic tool that facilitates the mimetic representation of non-verbal thought. It enables authors to effectively translate non-linguistic mental experience into language, playing an integral role of the type of ‘Stream of Consciousness’ narration that Chatman claims can “give a direct quotation of the mind – not merely of the language area but of the whole of consciousness” (1978: 187).

For the McEwan passage above, it is easy to demonstrate how the use of FIT reduces the implication of pure inner speech in the character’s thought content and thereby adds to the realism of mimetic thought presentation. If we change the discourse to the first person and present tense of FDT, we could plausibly imagine the character thinking the words ‘It is going well. It is going well.’ However, the more elaborate language that follows would seem over-articulate and reflective for plausible inner speech: ‘But for this one little thing I would be hugging myself, I would be dancing on the desk. It is rather like this morning when I lay in bed contemplating my success, denied full happiness by the single fact of Clive’s disapproval.’ With FIT and its partial narratorial grounding, the discourse cancels out the quotative effect of FDT, enabling the less pure but more naturalistic mimesis of thought content that does not necessarily consist of an unrealistic interior monologue of fully articulate inner speech.

Thus, by enabling narrative mimesis to delve beneath the surface-level cognition of inner speech and represent more profound, complex and realistic thought content, FIT creates some of the most interesting contexts for applying cognitive linguistics to the analysis of fictional minds. This, however, requires a more precise understanding of the different types of ‘conceptualizers’ whose mental activity constitutes the linguistic meaning. The next section will propose distinguishing three main types of conceptualizers according to the natures of their discourse involvement. It will lay the theoretical grounds for expanding the cognitive linguistic analysis of underlying mental activity to include both quoted characters and non-speaking characters in FIT situations. Finally, it will explain why the character’s mind (rather than the reader’s) might be the most interesting focus for this type of analysis.

5 Conceptualizer types in light of FIT

In light of the representational semantics of FIT, as well as the distinct participant roles that conceptualizers can take in fictional discourse that were discussed previously, I propose subdividing the concept of conceptualizer along similar lines to Chafe’s (1997) distinction between different types of ‘consciousness’ involved in discourse. This revised understanding of conceptualizers will create a clearer theoretical context for cognitive stylistics to analyze linguistic meaning directly in relation to the underlying mental activity of fictional minds (in addition to real minds), including non-speaking characters in FIT.

In spite of his central interest in ‘consciousness’ rather than cognition, Chafe shares ample territory with cognitive linguistics, especially Langacker’s CG. Understanding consciousness as ‘an active focusing on a small part of the conscious being’s self-centred model of the surrounding world’ (1997: 28), his main objectives are to describe how language reflects different ‘properties’ of consciousness, such as focus, dynamicity, and sources of experience (28–31), and to explain how consciousness ‘enters into’ discourse in different ways. For the latter issue, he deploys the concept of representation in a way that is compatible with the understanding of FIT above and which also helps to broaden the application of this concept to different types of discourse. Most importantly, Chafe’s distinction between representing and represented consciousness provides a basis for distinguishing between different types of conceptualizer, to be outlined subsequently, allowing us to reorient the analysis of linguistic meaning in cognitive stylistics.

Consciousness, according to Chafe, ‘enters into the production of language’ in two distinct ways (198). On the one hand, consciousness is responsible for producing language that represents ideas; and on the other, consciousness provides the ideas that are represented by language. On that basis, Chafe distinguishes between ‘a representing and a represented consciousness’ (198). In accordance with this distinction, there are two types of deictic elements depending on which consciousness they are oriented to. Tense and personal pronouns are oriented to the representing consciousness, whereas adverbs of space and time are oriented to the represented consciousness. Prototypically, the representing and represented consciousness are one and the same and both types of deictic elements are aligned. In that case, a speaker (the representing consciousness) produces language which represents their own ideas (the represented consciousness). Crucially, however, the two deictic systems pertaining to the two types of consciousness can be dissociated. In what Chafe calls ‘displaced immediacy’ (1997: 226), the representing consciousness can use language to represent the ideas of a represented consciousness other than their own. This is precisely what happens in FIT, which Chafe explains as language that is contrived by a narrator (the representing consciousness) to represent the consciousness of a character. This of course is perfectly in concordance with the understanding of FIT in the previous section.

Chafe’s delineation between represented and representing consciousness acknowledges the distinction between two properties inherent in our notion of subjectivity: agency and point of view. Discourse presumes these two properties in the forms of a subjective language producer on the one hand and the subjective semantic content expressed by language on the other. It is no coincidence that the deictic features aligning with these two properties of subjectivity are distinguished along precisely the same lines as deictic and grounding features aligned either with the narrator or the character in FIT. This is, after all, the most overt context in which the subjective act of discourse production is dissociated from the subjective ideational content it represents. I propose extending Chafe’s distinction from the more nebulous concepts of subjectivity and consciousness to the more tangible and analyzable domains of cognition and conceptualization as they are elucidated in cognitive linguistics. We can use it as the basis for identifying different types of conceptualizers involved in discourse, each of whose mental activity can be analyzed as linguistic meaning.

The typology of conceptualizers that I propose differs from Chafe’s typology of consciousness in two important ways. While Chafe makes his distinction along a single parameter – representing versus represented – I find it necessary to add a second parameter of intertextual versus extratextual, resulting in three main types of conceptualizers. Additionally, I find it important to include language receivers (e.g. hearers and readers) in addition to language producers within the conceptualizer types.

In the first type are what we can term extratextual conceptualizers. These exist independently of linguistic grounding and include anybody who happens to apprehend an expression through either production or reception. Language producers and their interlocutors fall into this category, but so do unacknowledged audiences (literary or otherwise) and even unintended audiences like eavesdroppers. It does not matter whether an expression is linguistically grounded in these individuals’ mental context; meaning can always be analyzed in relation to their conceptualization on the grounds that it is their minds that process that meaning in the real world.

The second type is the intratextual participant conceptualizers. These are the speakers and hearers in whose context an expression is linguistically grounded with tense and personal pronouns. This category includes not just prototypical speakers and hearers but also quoted speakers and their intended hearers, narrators and narratees, and also the inner speech of quoted thinkers. These conceptualizers are identified linguistically as participants involved in the discourse. Linguistic meaning can be analyzed in relation to their mental activity because their production or comprehension of the discourse, regardless of whether it is immediate or real, is intrinsic to its meaning.

The final type is the intratextual represented conceptualizers, in whose contexts the discourse is grounded as well, except in this case via deictic adverbs, aspect, modality, determiners, and evaluative and expressive language. These are the conceptualizers whose mental activity the language has been contrived to express. Again, prototypically speakers represent their own mental activity, but this category can extend to non-speaking characters whose ideation is represented for them with FIT.

To clarify the different types of conceptualizer it helps to emphasize that in prototypical discourse they are perfectly aligned. A speaker communicates with a hearer (the extratextual conceptualizers); the discourse is linguistically grounded in the speaker’s and hearer’s communicative context with tense and person pronouns (the intratextual participant conceptualizers), rather than being attributed to someone else; and it is also grounded in their contexts with deictic adverbs and other subjective features, purporting to represent their own conceptualizations (the intratextual represented conceptualizers), rather than those of a non-speaker. This alignment of conceptualizers breaks down as soon as the extratextual conceptualizer purports to quote the discourse of a different intratextual participant conceptualizer or to represent the ideation of a different intratextual represented conceptualizer.

Rather than continuing to elucidate the different conceptualizers in every different discourse situation (these should be fairly easy to extrapolate), we can concentrate on FIT where the breakdown of conceptualizer types is most complete and allows for a concrete set of options for analyzing meaning in terms of underlying mental activity. FIT is produced in the real world by an author and comprehended by a reader (the extratextual conceptualizers), but it is attributed to a fictional narrator (an intratextual participant conceptualizer) who uses it to represent the mental activity of a fictional character (the intratextual represented conceptualizer). A cognitive stylistician can justifiably analyze the language of FIT in relation to the underlying mental activity of any one of these conceptualizers. They could analyze how it reflects the author’s mental activity as they imagined the character’s mental activity while writing. They could analyze the fictional narrator’s mental activity in roughly the same way (the fictional narrator is not often explicitly differentiated from the author in FIT). In line with the dominant reader-response orientation of cognitive stylistics, they could analyze FIT in relation to the reader’s mind, determining how they use linguistic cues to ‘simulate’ the character’s mental activity in their imaginary experience of the fictional world (Nuttall 2018: 29). Or, in what seems to me to be the most natural and interesting focus for cognitive stylistics, FIT can be analyzed directly in terms of the character’s own mental activity, which is, after all, what the text is contrived and interpreted to represent.

A cognitive linguistic analysis that is directly oriented to the character’s represented mental activity allows stylistics to illuminate, in the most precise and direct way possible, the implicit understanding of the nature and workings of mind that are put forth in a fictional text. Of course, it is possible that the character’s underlying mental activities which are revealed through such an analysis also exist at some subliminal level in the mind of the reader; and it is even more likely that they existed in the mind of the author, assuming they carefully chose the language to best represent what they imagined the character’s mental experience to be like. Nevertheless, both author and reader are at best vicarious conceptualizers of a character’s conceptualizations, and in all probability they are unconscious of the cognitive processes that underlie this conceptual experience and are semantically attributable to the character’s mind. It is therefore at least justifiable, if not preferable, to apply cognitive linguistic analysis to FIT discourse (as well a character’s quoted discourse) in a way that backgrounds (or simply ignores) the language-processing activities of author and reader and concentrates solely on illuminating the mental activity that the discourse represents in the character’s mind.

6 Analysis

At this point it will be helpful to provide a brief example of the type of analysis being advocated here, applied to a passage from Charles Jackson’s The Lost Weekend (1944). It illuminates cognitive operations relating to framing and attention that are intrinsic to the semantics of the language and attributable to the protagonist’s mind.

The passage contains a combination of FIT and FDT, used to convey the alcoholic protagonist Don’s recall of his drunken attempt to steal a woman’s purse the previous night.

¹He went into the bathroom again, loosened his collar and tie, and began to shave. ²One thing they could never say about him: he was never not neat. ³He never let himself go to that extent. ⁴He still had some pride and self-respect left. ⁵But what good did it do him? ⁶Control was gradually slipping away. ⁷Who knew what he might not do next? ⁸I’ve got to watch myself, I’ve certainly got to watch myself; no telling what undreamed-of fantastic thing he might catch himself doing next. ⁹Catch himself? ¹⁰If he could only be sure of that! ¹¹But last night it had gone so far that others had caught him first (60).

The first sentence in this passage is the narration of events, but every sentence that follows is FIT, apart from two clauses of FDT in sentence 8. In FDT, Don is both the intratextual participant conceptualizer who produces the language in the fictional world and the intratextual represented conceptualizer whose ideation is expressed through the language (the ‘representing consciousness’ and the ‘represented consciousness’ in Chafe’s terminology). In FIT, he is only the intratextual represented conceptualizer; the intratextual participant conceptualizer for that discourse is the narrator. It is of course conceivable, even likely, that Don would produce at least some of this FIT language in the fictional world: if we translate it to the first person and present tense, as we did for the McEwan passage, for the most part it seems like more plausible inner speech. Nevertheless, the distal deictics create an important ambiguity in this regard, distancing the discourse from direct quotation and allowing for the possibility of at least some non-verbal thought. This semantic quality of FIT is amplified here by its direct juxtaposition with FDT, encouraging the treatment of the FIT thought content as something different than the fully articulate inner speech expressed with the other category (Cohn 1978). Nevertheless, in both FDT and FIT, it is Don’s conceptualization that is represented by the language; therefore, they can both be analyzed in terms of how they reflect his underlying mental activity.

We could analyze this underlying mental activity with cognitive linguistics in any number of ways, but one feature that stands out through sheer repetition is the use of a particular conceptual metaphor: ‘The Divided Person’ (Lakoff 1993, 1996). According to Lakoff, we have a folk model of our own psychology in which it is divided into two forces. The first force he calls ‘the Subject’, which is the locus of consciousness, perception, rationality and will; and the second he calls ‘the Self’, which consists of the body along with its irrational passions and needs (1996: 93). These are related to the concepts of the ‘superego’ and the ‘id’ in psychoanalysis (Talmy 2000: 460), and normally we desire the Subject to have control over the Self (Barreras Gomez 2015). The Divided Person metaphor manifests in several clauses in the passage, where Don is cast in dual semantic roles as both agent and patient in force dynamic relationships of ‘effective control’ (Langacker 2009: 153): ‘He never let himself go to that extent’; ‘I’ve got to watch myself, I’ve certainly got to watch myself’; ‘[…] no telling what undreamed-of fantastic thing he might catch himself doing next’; ‘Catch himself?’ The metaphor is also explicit in the concept of ‘self-respect’ and the idea of ‘[Self] Control […]slipping away’. And in the final sentence, the remark ‘others had caught him first’ plays on the same metaphor.

Don is thus shown to be heavily reliant on the Divided Person metaphor as a framing strategy to conceptualize his situation. He uses the concrete notion of two agents inclined to different courses of action to understand the more abstract situation of his alcoholism. One of these agents, his Self, is inclined to uncleanliness and bad behavior, while the other, his Subject, is obliged to monitor and control his Self. This conceptual metaphor is a tool he uses to make sense of a complex psychological situation that he does not understand very well, i.e. his alcoholism, by framing it in terms of more comprehensible interpersonal interactions rooted in force dynamic processes. This metaphorical framing would be likely to give him some reassuring sense of control over the problem, however illusory that sense may be.

We can expand this conceptual metaphor analysis by drawing on Langacker’s notion of ‘scope’ in CG (2008: 63), showing how Don’s alcohol consumption remains in the background of his conceptual experience without becoming a target of his attentional focus. Within any domain of conceptual activity that constitutes the meaning of an expression, CG distinguishes between ‘the immediate scope’ and ‘the maximal scope’ of attention. The immediate scope contains the portion of knowledge that is ‘foregrounded’ and directly relevant to the meaning of an expression, whereas the maximal scope contains knowledge that is backgrounded and indirectly relevant but nevertheless essential for the expression’s meaning (65). Information in the immediate scope is placed ‘onstage’ as a focus of the conceptualizer’s attention, while maximal scope information is ‘offstage’ and therefore not part of the attentional focus (77). The selection of immediate versus maximal scope inherent in an expression is the type of construal that Langacker calls ‘focusing’ (63).

In the above passage, immediate scope information relates to Don’s personal hygiene, diminishing self-control, ability to self-monitor, and the previous night’s misadventure, while backgrounded in the maximal scope is his reckless consumption of alcohol. For example, the essential contextual information while drinking is implicit in both the expressions ‘Who knew what he might do next?’ and ‘But last night it had gone so far’. However, the lack of explicit reference to this information means that it remains backgrounded, eluding his attention within his stream of consciousness. Returning to the Divided Person metaphor, the activity of the Self which the Subject strives to control would constitute the immediate scope information. In Don’s metaphorical thought, that controllable activity is limited to personal hygiene (‘… he was never not neat. He never let himself go to that extent’) and antisocial behavior (‘what undreamed of fantastic thing he might catch himself doing’). Once again, the real source of his problems, his alcohol abuse, is backgrounded in the maximal scope: his metaphorical conception places it offstage and outside the realm of the Subject’s control. Thus, underlying Don’s thought is both a conceptual metaphor which gives him a misleading sense of comprehension and control of his situation, and a self-deceptive attentional focus within that metaphor which evades confronting the alcoholism at the root of his problems.

This brief analysis serves to demonstrate how cognitive stylistics can treat a character as a conceptualizer by analyzing linguistic meaning directly in relation to the fictional mental activity that is represented with FIT and FDT (for a more extensive analysis in this vein, see Rundquist 2020). The main benefit of cognitive linguistics for this type of analysis is the way that it determines the semantics of the language specifically in terms of the conceptualizer’s underlying cognitive processes, providing a toolkit for illuminating them more directly and precisely. Of course, this cognitive stylistic methodology should be viewed as complementary to more traditional stylistic and narratological approaches.

A cognitive stylistic analysis of the above passage could also conceivably focus on the reader or author, i.e. the extratextual conceptualizers. In doing so, it would likely strive to explain how the language is processed in the mind of the reader or how it reflects the conceptions and intentions of the author, illuminating those real individuals’ mental activity in conceptualizing the narrative meaning. However, what is ultimately at issue in the meaning of a passage like this one is the protagonist’s mental activity. In recognizing that the character can be treated as a conceptualizer in his own right, it may ultimately be more fruitful for cognitive stylistics to by-pass reader and author and analyze linguistic meaning directly in relation to the character’s mental activity, which is what the language is contrived to represent. Nevertheless, the focus on any of these conceptualizers is a valid reflection of the CG understanding of meaning as conceptualization.

Finally, it should be noted once again that the character-focused analysis demonstrated here is not an entirely novel enterprise. Not only do older and more traditional mind-style analyses tend to focus on characters, some works in cognitive stylistics have, in parts, determined linguistic meaning in direct relation to character conceptualization (e.g. Giovanelli 2018; Semino and Swindlehurst 1996) and some even do so for passages of FIT (e.g. Nuttall 2018; Semino 2002, 2006). However, all the cognitive stylistic analyses are embedded within a more general focus on reader cognition, conforming to varying degrees with the dominant trend in this discipline to analyze character minds as by-products of reader experience. This article has elaborated the theoretical grounds in cognitive linguistics for treating character minds as objects of cognitive stylistic analysis in their own right.

7 Conclusions

In the introduction, I cited scholars who believe that literature can serve as a source of knowledge about the human mind and conscious experience; and in the subsequent sections I argued that a character-focused cognitive stylistic analysis is the best way to realize this potential. An important question remains unanswered; namely, why does literature have this potential in the first place? One reason is the exceptional transparency of fictional minds, which I discussed previously. However, that does not fully answer the question. Even if literature gives us deeper and more intimate access to the minds of others, why should we give any sort of scientific credence to fictional representations to inform our understanding of human experience? We might hypothesize that a reason we value many works of fiction has to do with the degree of detail, creativity and believability with which characters’ minds are represented. To achieve any of these qualities, authors must harness a certain degree of insight into how the mind works. They must engage in a sort of displaced introspection, imagining what an individual’s experience would be like in a certain situation and then representing that experience as best they can with the linguistic tools at their disposal. Fictional characters should be considered a valid attempt – perhaps humanity’s best attempt – to show what it is like to be somebody else.