Introduction and Overview

You have spent eight modules building an analytic apparatus that treats interview transcripts as texts to be coded. You have themes (Lesson 5), frameworks and conceptual models (Lesson 6), comparison engines (Lesson 7), and the counting machinery of content analysis (Lesson 8). That apparatus does an enormous amount of analytic work. It has one important limitation: it treats what participants say as the primary unit of evidence. Lesson 9 turns to the two traditions that ask a different question. Schema analysis asks what cultural and cognitive structures stand behind what participants say — the assumed-but-unspoken background that makes a particular utterance recognisable, normal, or transgressive. Narrative analysis (or narrative inquiry; see Clandinin & Connelly, 2000) asks how participants shape what they say into stories: the structures, sequences, evaluative pauses, and moral arcs by which a life is rendered tellable and a self is rendered coherent.

Bernard, Wutich, and Ryan (2017) devote two chapters to this material. Chapter 12 (pp. 269–287) is on schema analysis. Chapter 13 (pp. 288–304) is on narrative analysis. The chapters belong together because both ask the analyst to read for things that are not on the surface: a schema is the unspoken cultural model that organises what a participant takes for granted; a narrative structure is the unspoken story-shape that organises what a participant chooses to recount. In both cases the analytic move is interpretive in a different register from coding. You are not labelling segments. You are reconstructing what the participant is doing with language at a level the participant themself may not be able to articulate.

This section is about the first tradition. Schema analysis is the older of the two by several decades and is rooted in cognitive anthropology, an intellectual movement that asked, beginning in the 1950s, how culture lives in heads. The answer the field developed — that culture lives partly as schemas, learned mental structures that organise perception, memory, expectation, and action — turns out to be enormously useful for analysing what people say in interviews. When a participant says “you shouldn't be lonely at this age,” the analytical interest is not the surface sentence; it is the cultural schema of age-appropriate emotional life that the sentence is invoking, against which the participant is positioning themself. Schema analysis is the toolkit for surfacing those background structures.

1.1 The Intellectual Origins: Bartlett's Remembering

The schema concept enters the social sciences with Bartlett (1932) and his monograph Remembering: A Study in Experimental and Social Psychology. Bartlett, then at Cambridge, ran a now-famous series of experiments in which he asked English participants to read an unfamiliar Native American folk tale, “The War of the Ghosts,” and then to retell it from memory at intervals of minutes, days, weeks, months, and years. The retellings did not preserve the story. They reshaped it. Participants dropped culturally unfamiliar elements (the ghost-canoe), translated unfamiliar elements into familiar ones (canoes became boats, hunting seal became fishing), inferred missing causal connections, and progressively converged on a story that fitted English narrative conventions. Bartlett's interpretation: memory is not a tape recording. It is a reconstructive process that uses prior cognitive structures — he called them schemata — to fill gaps, normalise anomalies, and produce a coherent account.

For three decades after Bartlett, the schema concept lay relatively dormant in psychology while behaviourism dominated. It was revived by the cognitive revolution of the 1960s and 70s. Rumelhart (1980) gave the concept its modern computational form: schemas as data structures for representing generic concepts in memory, with slots that take particular values when filled by particular experiences. Schank and Abelson (1977) introduced the related concept of the script — a schema for a stereotyped sequence of events. Their canonical example is the “restaurant script”: enter, get seated, look at menu, order, eat, pay, leave. You do not consciously plan each step; you run the script. Scripts make the world navigable by encoding what to expect and what to do in routine settings.

Cognitive anthropology, particularly through the work of D'Andrade (1995) and The Development of Cognitive Anthropology, imported the schema idea from cognitive psychology and asked a culturally specific version of the question: which schemas do members of a particular cultural group share, and how do those shared schemas organise their behaviour, perception, and talk? Strauss (1992) and Quinn extended this line in A Cognitive Theory of Cultural Meaning, arguing that the relationship between culture and cognition is bidirectional. People internalise cultural schemas from social experience; the internalised schemas then organise individuals' perception and action, which reproduces (and gradually modifies) the cultural environment. Strauss and Quinn's monograph is the methodological foundation for contemporary schema analysis in anthropology and applied health research.

1.2 Schemas, Scripts, Prototypes, and Frames

Four cognitive structures are typically distinguished in this literature, and an analyst who is going to do credible schema work needs to keep them straight.

A schema is the most general term — a mental representation of a category or domain that organises perception and inference. The schema for “loneliness” in a given cultural group includes assumptions about who experiences it, under what conditions, how it feels, what it means, and what one ought to do about it. Two participants who share a schema do not necessarily share a belief about loneliness; they share a structure within which beliefs are generated.

A script (Schank & Abelson, 1977) is a schema for a temporally ordered sequence of actions and events. The “going to the doctor” script in mainstream Canadian healthcare runs: make appointment, arrive, check in, sit in waiting room, get called by name, get vitals taken in side room, wait in exam room, doctor enters, complaint described, examination, diagnosis or referral, leave. Deviations from this script (the doctor enters without knocking, the diagnosis is not given, the bill is presented in cash at the door) are noticeable precisely because they violate the script. Scripts are the analytic unit when you want to study how a culturally routinised activity is enacted and where it breaks.

A prototype (Rosch, 1973, 1978) is the best or most central example of a category — the exemplar against which other category members are judged. For most North American English speakers, “bird” has “robin” or “sparrow” as its prototype; ostrich and penguin are peripheral. Prototypes matter for schema analysis because participants reason about cases by comparison to a prototype, often implicitly. In loneliness research, the prototypical lonely person in mainstream English-speaking culture has been the elderly widow; that prototype shapes what counts as “real” loneliness and what counts as something else.

A frame (Goffman, 1974, Frame Analysis; Lakoff, 2004, Don't Think of an Elephant) is a structure that defines “what is going on here” in an interactional situation. Goffman's framing is interactional: the same scene (two people boxing) can be framed as a fight, a sport, a movie scene, or a rehearsal, and the framing determines the participants' understanding of what they are doing. Lakoff's framing is more politico-linguistic: the choice of metaphor (“the war on cancer,” “the journey with cancer”) frames the patient's experience and constrains downstream interpretation. Frames overlap with schemas but emphasise the active, often contestable, work of construing a situation.

1.3 Why Schemas Matter for Qualitative Health Research

The methodological pay-off of importing schema theory into health research is that it gives the analyst a vocabulary for what the participant is assuming. Most utterances in a transcript are not free-standing claims; they are responses to, elaborations of, or departures from a background structure that the participant takes for granted. Without a schema analysis, you can describe what was said. With a schema analysis, you can characterise what was presupposed — and what the participant therefore had to argue against, work around, or explicitly defend.

Consider three examples from health research where schema analysis has done substantive work. First, Strauss (1997) analysed how American working-class men talk about success and the meaning of life. She showed that the same speakers invoked two contradictory schemas — success-as-individual-achievement and success-as-providing-for-family — in ways that did not register as contradictions to the speakers themselves because the two schemas were active in different conversational contexts. Second, Kleinman's (1980, 1988) work on illness explanations across cultures showed that what counts as “the same illness” biomedically is parsed through profoundly different cultural schemas; analysing the schemas (not the diagnoses) is what reveals why patients comply or fail to comply with biomedical advice. Third, in tobacco-cessation research, qualitative work has documented a “smoker identity” schema (smoking as part of who I am) that competes with the “health behaviour” schema; interventions that address only the second schema fail when the first is operative.

For your loneliness capstone, the schema-analytic move opens up readings that pure coding cannot. When Frank (P20) says “loneliness is what nobody warns you about when you get old,” the surface content is a complaint about the inadequacy of warnings. The schema in the background is a cultural model of life-stage emotional preparation: there ought to be a warning, the speaker is entitled to one, the absence is a failure of someone (the culture? the generation before?) to prepare him. That schema — “emotional life-stages should be culturally announced and prepared for” — is invisible in a coded transcript that labels the utterance “loneliness in old age.” It is the analytic object of a schema reading.

1.4 Schema Analysis Within Cognitive Anthropology

Within the cognitive anthropology tradition, schema analysis is one of several techniques in a broader programme that includes cultural domain analysis, which you will meet more formally in Lesson 12 (Computational Text Analysis). Cultural domain analysis takes a bounded conceptual domain (kinds of illness, kinds of food, kinds of emotional state) and uses structured elicitation techniques — free listing, pile sorts, triadic comparisons, sentence frames — to map how members of a cultural group organise the domain. A free list of “kinds of feelings that are like loneliness” from twenty BC adults, analysed for frequency and salience (Smith's S), would give you a first map of the loneliness domain in this population. Pile sorts — asking participants to sort cards labelled with feeling terms into groups that go together — reveal the underlying conceptual structure of the domain. The full cultural-domain toolkit is covered in Bernard, Wutich, and Ryan's Chapter 16 and in Module 12 of this course.

For Lesson 9, the relevant move is the qualitative-text version of schema analysis, sometimes called discourse-based schema analysis. The analyst reads unstructured interview text for three signatures of an operative schema. First, normative assumptions: places where the participant treats something as the unmarked default (“normally you would think that...”, “most people just...”, “you should...”). Second, shared metaphors: recurring figurative framings across participants (loneliness as weight, as empty space, as a chair, as a closed door). Third, definitional moments: places where the participant has to do the work of saying what a category is, because the interlocutor or the participant themself does not take the meaning for granted. All three are visible in the loneliness transcripts and we will work with them in Section 2.

“A schema is what makes the unsaid sayable, the unremarkable noticeable, and the strange a little less strange. To analyse a schema is to make explicit the structure a participant has been navigating without naming.” — paraphrased from Strauss & Quinn (1997), A Cognitive Theory of Cultural Meaning.

Introduction and Overview

Section 1 made the case for schemas as analytic objects. This section is about how you get at them. There are two empirical strategies. The first is structured elicitation: you design instruments that surface schemas directly — free lists, pile sorts, sentence-completion frames, vignette responses, hypothetical situations. The second is discourse-based schema analysis: you read unstructured interview text the way schema analysts read it, looking for the three signatures introduced in Section 1 — normative assumptions, shared metaphors, and definitional moments. Bernard, Wutich, and Ryan (2017, Ch. 12) cover both strategies. Your loneliness capstone dataset is unstructured interview text, so this section weights the discourse-based work; but the elicitation techniques are introduced because they shape how the field as a whole studies cultural schemas and because you may want to add a structured elicitation to your own future research designs.

2.1 Structured Elicitation: Free Lists

A free list (Weller & Romney, 1988; Bernard, Wutich & Ryan, 2017, Ch. 16) asks participants to list, in any order, all the items that come to mind in a category. The instruction is something like: “Please list all the kinds of X you can think of, in any order. There are no right or wrong answers. Take your time.” Free lists are absurdly simple and analytically powerful. Two statistics are computed across participants: frequency (the proportion of participants who mentioned a given item) and average rank (where, on average, did the item appear in participants' lists). The two combine into Smith's S, a salience index that rewards items that are mentioned often AND mentioned early. The most salient items in a free list are the most central elements of the cultural schema for that domain.

For loneliness, a free list might ask: “Please list all the kinds of feelings that are related to or like loneliness.” Across thirty participants, you would expect “sad,” “isolated,” “empty,” “alone,” “disconnected” to appear with high frequency and early rank. Lower-frequency but still recurrent items (homesick, abandoned, invisible, restless) map the periphery of the schema. Items mentioned by only one or two participants are interesting in a different way: they may index sub-cultural variation, idiosyncratic experience, or genuine variability in the schema itself. The full machinery of free-list analysis is covered in Lesson 12 (Computational Text Analysis), but the principle — that salience reveals schema centrality — belongs in Lesson 9.

2.2 Structured Elicitation: Pile Sorts

A pile sort (Weller & Romney, 1988) gives participants a deck of cards with category labels and asks them to sort the cards into piles “that go together,” with no restriction on how many piles or how big each pile is. The analytic move is to compute, across participants, how often any two items end up in the same pile; the resulting similarity matrix can be visualised with multidimensional scaling or hierarchical clustering. The output is a map of the conceptual structure of the domain — which items are perceived as “close” to which others.

For a loneliness pile sort, you might prepare cards with twenty-five feeling-state labels (lonely, isolated, sad, depressed, homesick, abandoned, ignored, invisible, anxious, restless, bored, empty, longing, nostalgic, melancholic, blue, low, alone, secluded, withdrawn, disconnected, alienated, marginal, outside, unloved). Across thirty participants, you might find that “lonely” clusters tightly with “isolated” and “disconnected” for most participants but with “longing” and “nostalgic” for participants who experience loneliness as a temporal-relational phenomenon (missing someone) rather than a structural one (lacking ties). The pile sort is one way to surface that the cultural schema for loneliness is not unitary; it has variants that map to participants' life circumstances.

2.3 Structured Elicitation: Sentence Completion and Vignettes

A sentence-completion task gives participants a sentence stem and asks them to complete it. For loneliness research: “A truly lonely person is someone who...” or “Loneliness is worst when...” The completions surface the implicit definitional content of the schema. A high-frequency completion (e.g., “has no one to call”) is a load-bearing element of the schema; a low-frequency completion (e.g., “has stopped expecting connection”) may indicate a more elaborated or thoughtful schema. Sentence frames are particularly useful in cross-cultural work where direct translation of the category label is unreliable.

A vignette presents a short hypothetical scenario and asks participants to respond. For example: “Maria is 34, lives alone, has a job she likes, sees a few friends most weekends, and has not been in a romantic relationship for three years. Is Maria lonely? Why or why not? What might help?” Vignettes elicit schema-driven reasoning: participants invoke their schema of loneliness to judge whether the case counts as a member of the category, and the reasoning they articulate is the schema speaking. Vignettes are widely used in stigma research (Pescosolido, Monahan, Link, Stueve, & Kikuzawa, 1999) and in cross-cultural psychiatric epidemiology, where the goal is to see whether the same case is classified differently across cultural groups.

2.4 Discourse-Based Schema Analysis: Three Signatures

Most of your capstone work is on unstructured interview text. To analyse for schemas in such text, the discipline of the field is to read for three discourse-level signatures of an operative schema. The three signatures complement each other; any one alone is suggestive, but a credible schema claim is supported by all three.

Signature 1: Normative Assumptions

Look for places where a participant treats something as the unmarked default — the thing you are supposed to think, feel, or do, without having to defend it. Markers include modal verbs (should, shouldn't, ought to, have to), generic-statement framing (“most people just...”, “normally you would...”, “you'd think...”), and concessive constructions (“I know I'm supposed to...,” “they say you should...”). The participant is not stating their own view; they are invoking a cultural default and then often positioning themself in relation to it.

In the loneliness corpus, the most striking instance is Maya (P01), a 22-year-old undergraduate, who returns repeatedly to a normative assumption about life-stage and emotional state: “You shouldn't really be lonely at 22. That's not what 22 is supposed to look like.” She is not introducing this assumption; she is invoking and being constrained by it. The schema in play is something like: emotional life-stages are culturally announced, and 22 is the connection-rich stage. Maya's deviation from the schema is what makes her loneliness feel particularly disorienting to her.

Signature 2: Shared Metaphors

Look for recurring figurative framings across participants. Lakoff and Johnson (1980) and Metaphors We Live By argued that abstract domains (time, argument, emotion) are systematically structured by mappings from more concrete source domains (space, war, containers). When the same metaphor recurs across speakers, it is evidence of a shared cognitive structure — a schema — that is making the metaphor culturally available. The methodological move developed by Quinn (1991), in her analysis of American marriage talk, is to catalogue recurrent figurative expressions, group them by source domain, and read them for the schemas they enact.

In the loneliness corpus, several metaphor clusters recur. The most striking is the empty-space cluster. Frank (P20, 82, long-term care) says loneliness is “the empty space all those people used to fill.” Linda (P05, 67, widow) says loneliness is “Bill's chair” — a particular empty chair. Kenji (P14, 60, recently out) calls it “the absence of my children, even though they are alive and in the same city.” The shared structure is loneliness-as-absence, with a specific spatial-relational character: a place where someone used to be is empty now, and the emptiness is the loneliness. A second cluster is weight: Linda speaks of carrying “a weight,” Frank says “heavy, just heavy,” and several other participants describe loneliness as something that weighs them down or tires them out. The metaphor enacts a schema in which loneliness is a felt physical burden, not (or not only) a cognitive evaluation.

Signature 3: Definitional Moments

Look for places where the participant has to do the work of saying what something is. Definitional moments occur when the meaning of a category cannot be taken for granted — either because the interviewer has asked “what does that mean for you,” or because the participant senses ambiguity and rushes to clarify, or because the participant is articulating a distinction that mainstream culture does not. The definitional move is where the schema becomes available for inspection.

The richest definitional moment across the loneliness corpus is the recurrent distinction between alone and lonely. Almost every participant produces some version of it. Linda: “They've blurred together, I think. They didn't used to be the same.” Marcus (P08): “I'm rarely alone. I have lots of relationships. But I'm — culturally — there's a particular kind of absence...” Frank: “I'm rarely alone the way you might mean. But I'm lonely. Because the people here, they're strangers, mostly.” The participants are not introducing the alone/lonely distinction as a private insight; they are working a culturally available distinction and personalising it. The recurrence and the variation together are evidence of a shared schema with multiple permitted realisations.

2.5 Three Candidate Schemas in the Loneliness Corpus

Synthesising the three signatures across the twenty transcripts produces a working list of candidate cultural schemas. We will work with three of them.

2.6 Methodological Standards for a Credible Schema Claim

Schema claims are interpretive; they are not produced by counting code frequencies. The methodological discipline of the field, articulated across Strauss (1992), D'Andrade (1995), and Bernard, Wutich, & Ryan (2017), is to defend schema claims against four standards. First, specify the schema in propositional form: a one-sentence statement of the structure of expectation you are claiming. Second, provide multiple kinds of evidence: a credible schema is supported by at least two of the three discourse signatures (normative, metaphorical, definitional), across at least three participants. Third, look for and report counter-evidence: cases where the schema is absent, contested, or operating differently. Fourth, be explicit about the analytic level: a schema is a claim about cultural cognition, not about individual psychology. You are claiming that this structure is available to and recognised by members of the relevant cultural group; you are not claiming that every member endorses it.

Introduction and Overview

Schema analysis (Sections 1 and 2) reads the participant's talk for the cultural-cognitive structures it presupposes. Narrative analysis — the subject of this section and of Bernard, Wutich, and Ryan's Chapter 13 — reads the participant's talk for the story-shapes it produces. The two are complementary. The schema in a transcript is what the participant treats as given. The narrative in a transcript is what the participant assembles from given materials: a beginning, a middle, an end, a complication, an evaluation, a return.

The premise of narrative analysis is that narrative is a distinctive form of human cognition, not just a way of presenting information. Bruner (1991) argued that humans have two complementary modes of thought: the paradigmatic mode (logical, propositional, generalising, the mode of science) and the narrative mode (sequential, particular, evaluative, the mode of meaning-making). Both are universal; neither reduces to the other. Polkinghorne (1995) made the parallel methodological argument: the social sciences depend on narrative knowing as well as paradigmatic knowing, and the analytic neglect of narrative impoverishes our understanding of human experience. Riessman (2005), the most influential contemporary methodologist of narrative analysis, built on these arguments to develop a working taxonomy of narrative-analytic approaches — thematic, structural, dialogic/performative, and visual — that you should know by name even though this lesson focuses primarily on structural narrative analysis.

3.1 What Counts as a Narrative

Not everything a participant says is a narrative. In Labov's foundational formulation (Labov & Waletzky, 1967), a narrative is “one method of recapitulating past experience by matching a verbal sequence of clauses to the sequence of events which (it is inferred) actually occurred.” The minimum structural feature is at least two temporally ordered clauses whose order is constrained by the order of the events they recount — if you reverse the clauses, you change the story (“I left, then I cried” differs from “I cried, then I left”). The fully developed narrative has more than that, but the minimum is two ordered narrative clauses about something the speaker treats as a particular past event.

The distinction matters because much of what participants say in interviews is not narrative: it is description (“loneliness is like a heavy weight”), evaluation (“I think loneliness is worse than people admit”), generic statement (“I always feel it on Sundays”), and reasoning (“because the schema for what 22 should look like...”). All of these are analytically interesting and have been the subject of earlier lessons. Narrative is a particular thing — an account of a specific past event or sequence of events that the speaker is choosing to tell — and narrative analysis works on it as a distinct unit.

Most loneliness transcripts contain a small number of narratives, often elicited by Question 6 of the interview guide (“Can you walk me through a recent time when you felt lonely?”) and Question 15 (“Have you tried anything that didn't work?”). These elicitations explicitly ask for narrative. Other narratives appear unprompted, when a participant decides that telling a particular story is the best way to answer an evaluative or definitional question.

3.2 Labov's Six-Part Structural Model

William Labov, working as a sociolinguist in Harlem and Philadelphia in the 1960s and 70s, asked young men to tell stories about times they had been in physical danger. From the resulting corpus of personal-experience narratives, Labov and Waletzky (1967) derived a structural model that has become the standard for the close analysis of personal narrative. The model identifies six components, which may appear in canonical order or be re-shuffled, expanded, or compressed.

Of the six, evaluation is the analytically richest. Labov's insight was that personal narratives are not bare event sequences; they are told for a reason, and the evaluation component is where the teller tells the listener what kind of story this is and what it is supposed to mean. Evaluation can be densely concentrated in one place (an explicit “and that's when I realised...” line) or distributed across the narrative through subtle markers: comparison (“it was worse than the time before”), negation (“he didn't even look at me”), intensifiers (“I was so tired”), reported thought (“I thought, well, that's that”), and direct address to the interlocutor (“you know how it is”). Reading for evaluation is reading for what the narrator is doing with the story.

3.2.1 A Worked Example: Linda's FaceTime Birthday

Linda (P05) was asked to describe a recent time she felt lonely. Her response is one of the cleanest Labovian narratives in the corpus. Let us walk through it.

“Last Tuesday. My granddaughter's birthday. Michael's youngest. They're in Calgary, she turned six, and they did a FaceTime call, and it was lovely, it really was, but afterwards I — I just sat in the kitchen and I cried. Because Bill should have been there. He should have been singing happy birthday with me. And he wasn't. And then I made myself a cup of tea and I went to bed, but it was four in the afternoon. I was so tired. The grief and the loneliness, they make me so tired.” — P05, Linda, 67, retired teacher, widowed three years.

The narrative has all six Labovian components, in canonical order:

• Abstract: “Last Tuesday. My granddaughter's birthday.” Two terse fragments that announce what the story is about. Labov noted that abstracts are often very short and may be omitted altogether; here Linda gives one.
• Orientation: “Michael's youngest. They're in Calgary, she turned six, and they did a FaceTime call, and it was lovely, it really was.” The setting: who (Michael's youngest), where (Calgary, via FaceTime), what was happening (the call), and a first evaluation woven in (“it was lovely”).
• Complicating action: “...but afterwards I — I just sat in the kitchen and I cried.” The narrative shift. The “but” signals the turn from setting to event. The hesitation (“I — I just sat”) is doing evaluative work even mid-clause; this is the moment the loneliness arrived.
• Evaluation: “Because Bill should have been there. He should have been singing happy birthday with me. And he wasn't.” This is the heart of the narrative. The repeated should have been + the unmarked statement of fact (“he wasn't”) is Linda telling the listener what kind of story this is: it is a story about a particular absence, and the evaluation is the moral assessment that he ought to have been present.
• Resolution: “And then I made myself a cup of tea and I went to bed, but it was four in the afternoon.” The action that closes the event. The qualifying clause (“but it was four in the afternoon”) is another evaluation embedded in resolution: she is signalling that going to bed at four was a defeat.
• Coda: “I was so tired. The grief and the loneliness, they make me so tired.” The return to a generic present tense: this is what loneliness is, this is how it is now. The coda lifts the story out of its particularity into a general claim about her life.

The Labovian reading does work the simple coding does not. It identifies where in the narrative Linda is doing what kind of communicative work, and it lets the analyst see that the central evaluative line (“Bill should have been there”) is not just a piece of content but the structural pivot of the story — the place where Linda tells the listener what the story is about and why she has told it.

3.3 Plot Structures: Frye, Booker, and Genre Classification

Beneath the structural anatomy of any single narrative is a higher-level question: what kind of story is this? Northrop Frye (1957, Anatomy of Criticism) proposed that the entire Western narrative tradition organises around four fundamental mythoi or plot structures — comedy, romance, tragedy, and satire/irony — corresponding to the four seasons and to four trajectories of the hero. Christopher Booker (2004, The Seven Basic Plots) offered a related but more elaborate scheme. Riessman (2008) imports Frye into narrative analysis, arguing that participants tell their lives in identifiable plot structures and that recognising the genre is part of understanding the story.

• Romance: the hero leaves home, undergoes trials, and returns transformed. The plot of growth, of quest, of becoming. Many late-life-coming-out narratives are romances in this sense.
• Comedy: obstacles are overcome and the social order is restored or reconstituted. The plot of reconciliation, of marriage, of homecoming. Restitution narratives in chronic illness often have a comedic shape.
• Tragedy: the hero suffers a fall through fate or flaw; the social order is broken and not restored. Many bereavement narratives are tragedies; many narratives of progressive illness are tragedies.
• Satire/Irony: the narrative undermines its own meaning; expected resolutions fail; the world is shown as absurd or unjust. Common in narratives of structural marginalisation, of institutional failure, of betrayal by systems.

The point of the genre classification is not to reduce a complex narrative to a single label. It is to ask a question of the data: what shape is the participant trying to give to this experience, and what does that shape do for them? A narrative that is being told as a tragedy has different implications for intervention design (what kind of help would be wanted? what kind of help would be received as insulting?) than a narrative that is being told as a romance or a comedy.

3.4 Illness Narratives: Kleinman and Frank

The narrative analysis of health experience has a distinctive tradition that runs through two foundational books. Kleinman (1988) and The Illness Narratives: Suffering, Healing, and the Human Condition developed the methodological argument that biomedicine cannot understand chronic illness without understanding the narratives patients tell. Kleinman's earlier Patients and Healers in the Context of Culture had already distinguished three terms that the field now uses routinely: disease is the biomedical pathology, the thing diagnosed; illness is the patient's experience of the disease, the lived phenomenology; and sickness is the social and cultural meaning of the condition, the way it is categorised and responded to by family, employer, community, and state. Most biomedical encounters address disease; most patient suffering is in the illness and the sickness; and qualitative work on health narratives is where the illness and the sickness become legible. Greenhalgh and Hurwitz (1999) extended this argument into clinical practice, arguing that narrative-based medicine complements rather than competes with evidence-based medicine.

Frank (1995) and The Wounded Storyteller: Body, Illness, and Ethics is the second foundational text. Frank, a medical sociologist who had himself survived testicular cancer and a major heart attack, developed a tripartite typology of illness narratives that has become an analytic standard. His three narrative types — restitution, chaos, and quest — are not categories of illness but categories of story-shape, and most patients move between them over the course of their illness.

3.4.1 The Restitution Narrative

The restitution narrative is the dominant narrative of contemporary biomedical culture. Its plot is: I was healthy; then I got sick; I sought treatment; I am being treated; I will be healthy again. The hero is the patient, supported by medicine. The resolution is recovery. The arc is comedic in Frye's sense: the social order is broken and then restored. Restitution narratives are easy to tell and easy to receive because they fit the cultural script of biomedicine: there is a problem, a doctor, a treatment, a return.

Frank's critique of restitution as the default narrative is that it works well for acute, treatable conditions and works badly for chronic, progressive, or fatally ill conditions. Patients with chronic illness who try to maintain a restitution narrative when the disease will not co-operate are working against their own experience, and the narrative tradition that surrounds them often reinforces this. Listeners of restitution narratives want the patient to get better; they do not want the chaos that comes when restitution fails. Patients learn this and tell restitution narratives even when their own experience is not restitutional.

In the loneliness corpus, Tyler (P12, 29, recently dumped by his fiancée) tells a restitution narrative: he describes the loneliness as “a phase I'm getting through,” outlines a structured recovery plan (gym, therapist, new social commitments), and expects to be on the other side of it within a year. James (P02, 34, new father, postpartum-isolated husband) tells a similar restitution narrative: the loneliness is temporary, situational, and will end when the family routine settles. The narratives are coherent, future-oriented, and bounded.

3.4.2 The Chaos Narrative

The chaos narrative is, in Frank's terms, the antinarrative: the story of suffering that cannot be made into a story. The plot is something like I am sick; I am sicker; nothing helps; nothing makes sense; this is forever. Chaos narratives lack temporal structure, lack causal coherence, lack resolution; they are told in the present tense; they often cycle and repeat. Frank argues, controversially, that chaos narratives are hard to listen to and that the cultural pressure to tell a restitution or a quest narrative is partly the listener's discomfort with chaos.

Chaos narratives are common at the acute stage of severe illness, in serious depression, and in trauma. Methodologically, the chaos narrative is identified by its absence of the Labovian features of structured narrative: no clear orientation, no temporal sequence, no resolution, often no abstract. The reader of a chaos narrative cannot easily extract a story; the story is the lack of story.

In the loneliness corpus, Diana (P07, 41, working-mother burnout, late-stage divorce) and Elena (P16, 47, long-COVID symptoms) tell narratives that approach chaos. Diana's account of her loneliness moves in circles — she lists what she has tried, says nothing has worked, lists more things she might try, says she doesn't have the energy, returns to the same starting point. There is no through-line. Elena describes her current state as “everything is broken at once and I can't figure out which thing to put down first.” The analytic move is to read these accounts as chaos narratives rather than as failures to communicate — the chaos is the communication.

3.4.3 The Quest Narrative

The quest narrative is the story of illness as a journey that has changed the teller. The plot is: I was who I was; the illness happened; I have been transformed; I am now someone different, possibly someone better. Frank distinguishes three subtypes — the memoir (illness as an event that has been integrated), the manifesto (illness as a call to social change), and the automythology (illness as a heroic ordeal that has produced a new self) — but the common feature is transformation. The arc is romantic in Frye's sense: the hero leaves home, undergoes trials, returns transformed.

Quest narratives are widely admired in our culture and have a kind of moral prestige that restitution and chaos narratives do not. The risk is twofold: patients can feel pressure to construct quest narratives prematurely, before the illness has actually permitted transformation; and listeners may receive a quest narrative as more authentic than a chaos narrative, which is unfair to people whose experience is genuinely chaotic.

In the loneliness corpus, Kenji (P14, late-life coming out), Linda (P05, late-life widow), and Rose (P19, neurodivergent late-diagnosed adult) each tell quest narratives. Kenji's is the clearest: the loneliness for his sons is paired with a transformation (new partner, tea ceremony, recovery of Japanese heritage) that he is explicit about. He is becoming someone different, and the becoming is part of how he understands the loneliness. Linda's quest is quieter — the loneliness is “the cost of love” and she is learning to make peace with it — but the structure is the same: an event has happened, and she is being changed by her engagement with it.

3.5 Narrator-Self, Narrated-Self, and Narrative Identity (McAdams)

One more distinction is essential for credible narrative analysis. The narrated-self is the person inside the story: the past Linda sitting in the kitchen crying after the FaceTime call. The narrator-self is the person telling the story now: the present Linda speaking to the interviewer in her living room. The two are not the same. The narrator-self has access to the outcome the narrated-self did not have. The narrator-self chooses what to include, what to elaborate, what to leave out. Evaluative moments — especially external evaluation, where the narrator steps outside the story to comment — are where the narrator-self is most visible.

McAdams (1995), in his life-story theory of personality, proposed that adult identity is itself a narrative construction: people develop a narrative identity, an internalised and evolving life story that integrates the reconstructed past, perceived present, and anticipated future, providing a sense of unity and purpose. From this perspective, every personal narrative is a moment of identity work: the narrator-self is using the story to construct or reinforce a particular version of themselves. McAdams' large empirical corpus showed that mid-life North Americans disproportionately tell redemption sequences (bad event → good outcome) and that the cultural prestige of redemption shapes individual life-story construction.

For your loneliness capstone, the narrator-self / narrated-self distinction is the discipline that prevents you from collapsing the participant into the story. Linda the narrator-self is composed, articulate, reflective — she gives a coda about the cost of love. Linda the narrated-self in the kitchen is none of those things; she is shattered. The narrative analysis attends to both and to the gap between them: what kind of narrator-self is Linda constructing through this telling, and what is it costing her or giving her to construct it?

“Narrative is the means by which existence is made sensible and bearable. When the narrative breaks down — chaos — existence becomes unbearable. When a new narrative is constructed — quest — existence is recovered on changed terms.” — paraphrased from Frank (1995), The Wounded Storyteller.

Introduction and Overview

Sections 1, 2, and 3 introduced schema analysis and narrative analysis as parallel interpretive techniques. This section applies them to the loneliness corpus in detail, sets up the Week 9 capstone milestone, and demonstrates a light R + Taguette workflow for tagging narrative fragments. The intent of this section is operational: by the end, you should know which transcripts to read for which kind of analysis, what a defensible schema or narrative memo looks like, and how to organise the qualitative work so that the analytic claims you eventually make in the capstone are defensible.

4.1 Frank's Typology Applied: Restitution, Chaos, Quest Across the Corpus

The loneliness corpus distributes unevenly across Frank's three narrative types. Several participants tell predominantly restitution narratives, several tell quest narratives, and a smaller number tell narratives that approach chaos. The mapping below is a working analytic classification — not a final claim — and the methodological discipline of the field requires that each classification be supported by evidence and that competing readings be acknowledged.

Three methodological points fall out of this mapping. First, participants frequently mix narrative types within a single interview. Linda's interview contains quest elements (the cost-of-love framing) and restitution elements (the routines that “keep me alive”) and momentary chaos (the four-PM bedtime). The classification is of the dominant register, not the totality. Second, the classification is not a value judgement. Quest narratives are not better than chaos narratives. They are different shapes that experience can take, and the political-aesthetic preference for quest narratives in our culture is itself a finding that the careful analyst notes. Third, the classification has analytic consequences. A restitution narrative invites a particular kind of intervention (support during the bounded period); a chaos narrative invites a different kind (presence and witnessing without restoration-pressure); a quest narrative invites yet another (recognition of transformation, scaffolding for the new self). Public-health interventions calibrated to the wrong narrative type are unlikely to land.

4.2 An Extended Labovian Reading: Kenji's Birthday Card

The worked Labovian example in Section 3 was Linda's FaceTime story. Here is a second, from Kenji (P14), illustrating how Labov's six components work in a narrative whose evaluative line is more diffuse.

“My older son's birthday last month. I sent a card and a gift. He returned the gift. The card came back marked 'return to sender'. And I sat with that for, you know, several days. My partner was very kind. But there is no one who can take the place of my son not wanting to know me. So the loneliness was very acute that week.” — P14, Kenji, 60, retired civil engineer, recently out, estranged from his sons.

• Abstract: “My older son's birthday last month.” Compressed announcement of topic.
• Orientation: implicit, carried by knowledge of the participant's earlier disclosure that his sons are not speaking to him; minimal scene-setting in this fragment.
• Complicating action: Three terse event clauses in tight temporal sequence: “I sent a card and a gift. He returned the gift. The card came back marked 'return to sender'.”
• Evaluation: Distributed and largely external. The first evaluative move is the pause: “And I sat with that for, you know, several days.” The hedging (“you know”) and the temporal stretch (“several days”) is doing the evaluative work of saying this was the kind of event one has to sit with. The second evaluative move is the gnomic statement: “there is no one who can take the place of my son not wanting to know me.” This is external evaluation in the strict Labovian sense — Kenji steps out of the story to comment on it.
• Resolution: The resolution is the evaluation; there is no further event. This is structurally common in narratives of irresolvable grief.
• Coda: “So the loneliness was very acute that week.” Lifts the story back to the interview-level framing.

The Labovian reading reveals two things. First, the resolution-as-evaluation structure is itself a finding: Kenji's narrative ends not with a resolving action but with a moral statement, which signals that this is a kind of story where action does not resolve. Second, the gap between narrator-self and narrated-self is small in this fragment — the present Kenji is still in the situation the past Kenji was in. The lack of resolution and the small narrator/narrated gap together suggest a narrative that is still being lived, which has implications for what kind of witnessing, intervention, or support would be coherent with the participant's actual situation.

4.3 Combining Schema and Narrative Readings

The two analytic approaches are complementary and most powerful in combination. A schema reading of Kenji's account would identify the operative schema as something like a father is owed his children's recognition; rejection by a child is a particular kind of catastrophe. A narrative reading would identify the story-shape as a quest in progress with chaos episodes. The two readings together do work neither does alone: the schema tells you what cultural structure the narrator is moving within; the narrative tells you what shape he is trying to give to his movement; and the combination tells you both the constraint and the agency in the participant's account.

For your capstone, you do not have to do both readings. The Week 9 milestone (below) asks you to choose. But a sophisticated full capstone paper might use schema analysis to characterise the cultural context against which participants are working and narrative analysis to characterise what individuals do within that context. The two analyses then occupy adjacent paragraphs in the findings section.

4.4 R Workflow: Extracting and Tagging Narrative Fragments from Taguette

The R toolchain you set up in Lesson 1 supports a light-touch workflow for narrative analysis. The key move is to use Taguette to identify and tag narrative fragments by Frank typology during your hand-coding pass, export the tagged extracts to a CSV, and bring them into R as a tidy dataframe for cross-case comparison.

library(tidyverse)
library(stringr)

# Read the Taguette highlights export
narratives <- read_csv("taguette_loneliness_highlights.csv") %>%
  filter(str_starts(tag, "narrative_")) %>%
  mutate(
    frank_type    = str_remove(tag, "narrative_"),
    participant   = str_extract(document, "P\\d{2}_[A-Za-z]+"),
    fragment_chars = nchar(content)
  ) %>%
  select(participant, frank_type, content, fragment_chars)

# Cross-case summary: how many fragments of each type per participant?
narratives %>%
  count(participant, frank_type) %>%
  pivot_wider(names_from = frank_type, values_from = n, values_fill = 0) %>%
  arrange(desc(quest + restitution + chaos))

# A working “dominant type” column for each participant
participant_dominant <- narratives %>%
  count(participant, frank_type) %>%
  group_by(participant) %>%
  slice_max(n, n = 1, with_ties = FALSE) %>%
  ungroup() %>%
  rename(dominant_type = frank_type)

# Pull all quest fragments for close re-reading
quest_fragments <- narratives %>%
  filter(frank_type == "quest") %>%
  arrange(participant)

writeLines(paste0("--- ", quest_fragments$participant, " ---\n",
                quest_fragments$content, "\n"))

Two points about this workflow. First, R is doing administrative work, not analytic work. The Frank-typology classification is yours, made by hand in Taguette as you read. R organises the resulting tagged fragments so you can see them across participants. Second, the workflow scales. If you tag thirty fragments across eight participants, R will print them in a useful form. If you tag three hundred fragments across twenty participants, R will still print them in a useful form. The same workflow is what published narrative-analytic health studies use to manage their fragment-level data.

4.5 The Week 9 Capstone Milestone

Bringing It All Together

Lesson 9 has done a particular kind of work in the arc of HSCI 841. Lessons 5 through 8 built the engineering of qualitative analysis: themes, codebooks, frameworks, comparison, and content analysis. Lesson 9 turns to the interpretive registers that this engineering supports but does not constitute. Schema analysis (Sections 1–2) reads for the cultural-cognitive structures that participants presuppose; narrative analysis (Section 3) reads for the story-shapes participants produce; and Section 4 applied both to the loneliness corpus. Together, the two techniques give you analytic access to the layer of qualitative data that lies between coded content and surface meaning — the layer where culture lives in participants' talk and where participants do identity work through their stories.

The Week 9 capstone milestone — an 800–1000-word memo applying either schema or narrative analysis to a subset of transcripts, with quoted evidence and a structural diagram — is your first piece of interpretive-register analysis in this course. The memo will be referenced again in your final capstone paper, either as a stand-alone finding or as a piece of analytic context for the comparative work you will do in later lessons.

The substantive scholarly inheritance you are working with is large. Schema analysis sits in a hundred years of cognitive anthropology, from Bartlett through D'Andrade to Strauss and Quinn. Narrative analysis sits in the “narrative turn” of Bruner, Polkinghorne, and Riessman, with the foundational illness-narrative work of Kleinman and Frank as its medical-humanities anchor, and the structural toolkit of Labov as its close-reading instrument. You do not need to master all of this to do credible work, but the more of it you can recognise by name and use accurately, the more your eventual paper will speak the language of the field.

The final reflection below asks you to articulate what schema or narrative analysis adds to your analytic toolkit that the methods of Lessons 5–8 do not. The submit button on the final assessment will not unlock until you complete this reflection (minimum 30 characters).

This glossary collects the key concepts, people, and methodological stances introduced in Lesson 9. Use it as a reference while you work through the material, or as a review before the final assessment. Type in the search box to filter entries.