# Lesson 9: Schema and Narrative Analysis

*Companion-podcast transcript • Sarah & Kiffer*  
*Office Hours episode to listen to after working through the lesson*

---

**Sarah:** Here's a sentence from the course's loneliness interviews. Maya, who's twenty-two, says, "You shouldn't really be lonely at twenty-two. That's not what twenty-two is supposed to look like." So what has she just told us?

**Kiffer:** At least two things. She's told us how she feels, and she's told us a rule about how a life is supposed to go. She didn't write that rule, and she can't seem to get out from under it. That rule is what a schema analyst is after.

**Sarah:** Or she's told us a third thing. She's talking to a stranger with a recorder, and she's letting that stranger know she understands this isn't normal, before the stranger has a chance to think it.

**Kiffer:** And whether you can tell those readings apart from a transcript alone is one of the real arguments about schema analysis.

**Sarah:** Welcome back to Office Hours. I'm Sarah.

**Kiffer:** And I'm Kiffer. Maya's sentence sits right at the centre of Lesson nine, Schema and Narrative Analysis, so let's start there.

**Sarah:** Question one. What makes something a schema, and how is it different from a theme or a belief? When people first try schema analysis, they often find a recurring topic, call it a schema, and move on.

**Kiffer:** Maya is a good case for pulling the three apart. The theme is the recurring topic, something like age and loneliness, and you could find it in a dozen transcripts and count it. A belief is a proposition a person can state and would sign their name to. Now ask Maya whether she believes twenty-two-year-olds are never lonely, and she'd probably say of course some are. She's one of them.

**Sarah:** So the schema is operating even though she'd reject it if you put it to her as a belief.

**Kiffer:** That's the heart of it. A schema is a structure of expectation working in the background. It fills in what's normal, it makes a departure feel like a departure, and you recognize it by its effects on talk. That's why you write it as one proposition and then defend it with more than one kind of evidence.

**Sarah:** Then let's build one. Here are three invented lines from three made-up participants in a hypothetical loneliness study. The first says, "I'd never call someone just to say I'm lonely. You call when you've got news, or when somebody's sick." The second says, "My mom always said, don't be a burden. So I text first, and I always add, no pressure if you're busy." The third says, "Reaching out is like asking for a loan. You can only do it so many times before you owe people."

**Kiffer:** So here's the question for everyone listening. What's the schema behind these three lines, stated as a single proposition? Take a few seconds. Pause the audio if you'd like more time.

*(Pause)*

**Sarah:** I'd write the proposition like this. Contact needs a reason, and asking for company without one puts you in someone's debt. And for the signatures, the first line is a normative assumption, a generic statement about when you're allowed to call. The second passes on a rule, don't be a burden, and burden is also a metaphor. The third is a metaphor from money.

**Kiffer:** That's a good statement, and notice that nobody in the data said it. You reconstructed it from three people who each supplied a piece of it, and you've read the signatures correctly.

**Sarah:** The two metaphors are different, though. A burden and a loan. Does that still count as a shared metaphor?

**Kiffer:** Good question. What they share is the source idea that your need costs the other person something. Naomi Quinn found this pattern in her work on how Americans talk about marriage. People reached for quite different metaphors, and the metaphors kept expressing the same few underlying ideas. So you catalogue the metaphors by their source, then read across the sources for the idea they carry, and the idea is where the schema shows itself.

**Sarah:** Then on the standards, we have one proposition and two kinds of evidence across three people. We still need counter-evidence and the level.

**Kiffer:** Right. For counter-evidence, look for people who call a friend for no reason and treat it as ordinary, or who describe being asked for company as a gift. For the level, say plainly that this is a cultural model these speakers can draw on. The claim is about a model that's available to them, and it leaves open whether any one of them personally holds it.

**Sarah:** And if you wanted to test it beyond interview talk, a vignette would do it. Something like, "Sam hasn't spoken to anyone in a week and texts a friend just to chat. What does the friend think?" If people reason about Sam in terms of owing, the schema is showing up in a second kind of data.

**Kiffer:** That's a nice design, and it shows why the schema matters for public health. Picture a campaign built on the script from the lesson, reach out to a friend when you're struggling. For people who carry the debt schema, that message asks them to take out a loan. Knowing the schema suggests changing the message, perhaps by framing contact as something you give.

**Sarah:** Question two. What counts as a narrative, and where do you find the evaluation? I'd like to try this one myself, because Labov's model looks easy until you sit down with a real story.

**Kiffer:** Then here's an invented story from a hypothetical participant, a man in his fifties who moved to a new city for work. "So, Canada Day this year. I'd been in the city maybe four months. Back home I used to go to the fireworks every year with my brother. Everyone at work had plans. I walked down to the waterfront by myself, and I stood there for the whole show, and nobody said a word to me. Not one person. I mean, why would they? Then I walked home and put the TV on. That's sort of how holidays go now."

**Sarah:** Before I try it, a question for everyone. Which clauses are the complicating action, the spine of what actually happened? Take a few seconds.

*(Pause)*

**Sarah:** Here's my attempt. The abstract is "So, Canada Day this year." The orientation is the four months in the city, the fireworks with his brother back home, and everyone at work having plans. The complicating action is that he walked to the waterfront, he stood there for the whole show, and nobody said a word to him. "Not one person" and "why would they" are evaluation. The resolution is walking home and putting the TV on, and the coda is "that's sort of how holidays go now."

**Kiffer:** Most of that holds up. Let me push on one clause. Try the swap test from the lesson on "nobody said a word to me." What event does that clause report?

**Sarah:** Okay. If I swap it with "I stood there for the whole show," the story doesn't change at all. And nobody saying anything is, well, nothing happening. There's no event there.

**Kiffer:** So what is that clause doing in the story, then?

**Sarah:** It's comparing what happened with what could have happened. In a crowd on Canada Day, somebody might have said something. So it's evaluation, a negative. And it's actually the most important evaluation in the story. Take it out, and you've got a man who watched fireworks and went home, and most of the reason to tell it is gone.

**Kiffer:** That's exactly the correction, and it matches Labov's own treatment. Negatives set the events that happened beside an alternative that didn't, which is why negation sits among the evaluation markers. Notice "by myself" too, tucked inside an action clause. The evaluation is spread through this story, and the strongest piece of it is a non-event.

**Sarah:** What about the line about going to the fireworks with his brother? I put it in the orientation.

**Kiffer:** That's defensible. "Used to" describes something that happened many times, so it can't be ordered with the events of that one night, and that's why it isn't a narrative clause. It does two jobs. It sets the scene, and it sets up the comparison that makes this year hurt. Many analysts would call it orientation with evaluative force.

**Sarah:** And the coda, "that's sort of how holidays go now," is the narrating self speaking.

**Kiffer:** Yes. Today's narrator is turning one evening into a general claim about his life. The "sort of" softens it, maybe because a flat statement could sound like self-pity to a stranger. The man on the waterfront was living one night. The man in the interview is deciding what that night says about him, and that gap is where the identity work happens.

**Sarah:** It's also a reminder that most of an interview isn't narrative at all. "I always feel it on holidays" is a generic statement. It might lead into a story, but on its own it has no events in order.

**Kiffer:** Right, and that matters when you go looking. Narratives tend to cluster where the guide asks for them, like a question about a recent time someone felt lonely. Others arrive unprompted, when a participant decides a story is the best answer to an abstract question, and those are often worth the closest look.

**Sarah:** Question three. Frank's restitution, chaos and quest types come from illness narratives. Loneliness isn't an illness. Are we entitled to use them?

**Kiffer:** It's a fair challenge. Arthur Frank, a sociologist who taught for decades at the University of Calgary, built the types from stories of serious illness and set them out in The Wounded Storyteller in 1995. Kleinman's terms help sort it out. Loneliness clearly has an illness layer, the lived experience, and a sickness layer, its social meaning, and interviews are full of both. What it lacks is a settled disease layer, something a clinician diagnoses in the body.

**Sarah:** Although public talk is moving that way. In 2023 the United States Surgeon General published an advisory called Our Epidemic of Loneliness and Isolation.

**Kiffer:** And that title is a frame in Lakoff's sense. "Epidemic" brings a disease frame along with it, and a disease frame makes the restitution story easy to tell. You got hit, you got help, you recovered. Tyler's plan in the loneliness interviews, with the gym, the therapist and new social commitments, reads a lot like a treatment plan. So when you classify a loneliness story as restitution, it's worth asking whether the participant is using a medical script that public talk about loneliness is handing them.

**Sarah:** So the types can travel to loneliness, as long as we remember they describe the shape of a story, and the shapes available to people come partly from how loneliness is talked about in public.

**Kiffer:** That's how I'd put it. Let's try an example. Here are three invented fragments from one hypothetical participant, a nurse in her thirties who took a job in a small town. First: "Once the night shifts end in the spring, I'll have weekends again, and I'm going to join the curling club." Second: "Some weeks it's the shifts, and then the car, and then my sister doesn't pick up, and then I'm just sitting in the parking lot, and I don't know." Third, from the very end of the interview: "I think I've learned I need people more than I ever let myself admit." Most of her interview sounds like the first fragment, with stretches like the second.

**Sarah:** So, for listeners. Which of Frank's three types does each fragment lean toward? Take a few seconds.

*(Pause)*

**Sarah:** The first leans restitution. It's bounded, it has a date, and it expects a return to normal. The second leans toward chaos. It's in the present tense, it strings things together with "and then" with no sense of cause or direction, and it trails off. The third sounds like quest, a changed self. I'd be careful with that one, though. It's a single line at the very end, the narrating self summing everything up, and most of what she actually told us sounds like the other two.

**Kiffer:** That's a good catch. Frank described chaos as time without sequence, and he heard that same "and then, and then" rhythm in chaos stories. Here it piles up troubles in the present tense, which is a different job from the "and then" that moves a Labov story from one past event to the next. For the dominant register, I'd call it restitution with chaos breaking through, report the quest line as a competing reading, and say where it sits in the account. And notice why it's tempting to promote the quest line. It's the tidiest, most hopeful sentence in the interview.

**Sarah:** That's Bartlett's finding from the lesson. His English participants retold an unfamiliar folk tale, and in some retellings the canoes turned into boats. Analysts retell participants' stories too, and a chaos fragment can quietly become a quest when it's written up.

**Kiffer:** Which is why reading chaos as chaos is a discipline for the analyst as much as for the clinician. If your memo tells a neater story than the transcript does, it's worth asking whose schema did the tidying.

**Sarah:** I want to push on one more idea from the lesson, that support pitched at the wrong narrative type tends to miss. Does that mean a service should work out which type someone's in and match the help to it?

**Kiffer:** I'd be cautious about that. The types describe an account at a moment, and this nurse moves between at least two of them inside one interview. I think the defensible use is as a way of listening. If someone's in chaos, you hear it as chaos before offering a plan. I wouldn't turn the typology into a label that sits in someone's file.

**Sarah:** Question four. How do you know when you have enough evidence for a schema claim? The standard from the lesson is at least two of the three signatures across at least three participants. I'd like to test it on something harder.

**Kiffer:** Then here's an invented memo to appraise. A hypothetical team interviews eight people caring for a family member with dementia, and asks about loneliness. Their memo claims a schema: loneliness is a weight you carry alone. The evidence is a weight metaphor in six of the eight transcripts, plus generic statements like "that's just how it is now" from five participants, which the team codes as normative assumptions. Two signatures, more than three people. Before you'd accept that, look again at those generic lines. What else could they be? Take a few seconds.

*(Pause)*

**Sarah:** They sound a lot like the end of the Canada Day story. "That's sort of how holidays go now." If those lines close off stories, they could be codas.

**Kiffer:** Good. Suppose you check, and all five turn out to be the last line of a story about one particular evening or visit. One caregiver describes visiting her father in long-term care and finishes with, "So that's just how it is now."

**Sarah:** Then they're doing what a coda does. They bring a story back to the present and turn it into a general claim about the speaker's own life. A normative assumption invokes what people in general are supposed to feel or do, and none of these five lines reaches for that. So the normative evidence mostly disappears.

**Kiffer:** That's the right reclassification. You could argue the "just" in "that's just how it is" hints at a default, so I'd call these lines weak evidence at most. They can't carry a signature on their own. And the metaphor has its own problem. English describes a lot of hard feelings as heavy. Grief is heavy, worry is heavy, a long month is heavy.

**Sarah:** So a metaphor tells you about this schema when it carries the schema's idea, the part about carrying it alone. That's Quinn's point again, that you read a metaphor for the idea it's used to express.

**Kiffer:** Right. Suppose that in four of the six transcripts the word is simply heavy, and only two extend it into something specific, like "there's nobody to hand it to." Those two are the ones that speak to carrying it alone.

**Sarah:** Let me count what's left. The specific version of the metaphor from two people. Normative evidence too weak to count once the codas are reclassified. No definitional moments reported. So it's one signature across two people, and it fails the standard it seemed to pass. What would fix it?

**Kiffer:** Three things. Look for the schema away from story endings, in places where people talk about what caregivers are supposed to feel, something like "you're not supposed to complain when it's your own dad." Look for definitional moments, where someone says what this loneliness is. And take the codas over to the narrative side, because five people closing their stories in the same resigned way is a real finding about how these stories are told.

**Sarah:** That's a trap a frequency count would never catch, because the counts were real. Six of the eight really did talk about weight, and five really did say something like "that's how it is now."

**Kiffer:** Yes, the counting was accurate, and the inference was wrong. The same line can be a coda in a narrative reading and look like a normative assumption in a schema reading, so before you count a line as evidence, work out what job it's doing in the telling.

**Sarah:** Okay. Now the part where we disagree, and I'm going back to Maya.

**Kiffer:** Go ahead, and make the strongest case you can.

**Sarah:** Schema analysis reads "you shouldn't really be lonely at twenty-two" as a window onto a cultural model that people carry around in their heads. An interview is a conversation, though. Maya is admitting something stigmatized to a stranger. Saying she knows it isn't normal shows the listener she's a reasonable person who knows the rules. That's an action she's performing in the talk, and I don't think you can read a mental structure off it.

**Kiffer:** I agree with the first half. It is an action. Where I'd differ is on what follows. Getting ahead of a judgement only works if the listener would recognize the judgement. Maya can count on a stranger knowing that twenty-two is supposed to be a connected time of life. That shared expectation is exactly what the schema claim is about.

**Sarah:** So you'd give up any claim about what's in her head.

**Kiffer:** I wouldn't make claims about her head from one sentence. The fourth standard from the lesson already sets the level as cultural: the model is available to speakers and recognizable to listeners. Her self-presentation is evidence for that, if anything.

**Sarah:** But then you have to accept that the "should" might be there because of who she's talking to. With a friend her own age she might never say it. The interview produces some of the normative talk you're treating as evidence.

**Kiffer:** That's true, and it's the strongest form of your argument. A transcript shows you the schema as it appears in one kind of conversation, with a researcher, about a stigmatized feeling. I'd still count that as real evidence of a shared model. I'd be more careful about how much it tells you about the rest of people's lives.

**Sarah:** I'm more skeptical than Kiffer about how far into cognition this reaches, but we agree on the practical rule. Read every "should" as a move in the conversation it belongs to, and state the schema as a model speakers can count on listeners to recognize.

**Kiffer:** Let's pull it together with three things to take away.

**Sarah:** First, a schema is reconstructed. You write it as one proposition that usually nobody in the data says outright, and you defend it with more than one kind of evidence, with counter-evidence, and with a clear statement that the claim is cultural.

**Kiffer:** Second, in a narrative, evaluation often hides in the non-events. A line like "nobody said a word to me" can be what makes a story worth telling, so read for what didn't happen as well as what did.

**Sarah:** And third, work out what job a line is doing before you count it. A closing line like "that's how it is now" may be a coda, a word like heavy may be ordinary English, and a hopeful line at the end may be the narrating self summing up.

**Kiffer:** If you'd like more practice, take the Canada Day story from this episode and write out your own Labov reading. Then change one line, so that somebody nods at him during the fireworks, and notice how much of the evaluation disappears.

**Sarah:** Next time, it's Lesson ten, Discourse Analysis: Conversation and Performance, where the transcript becomes a record of what people do with talk, turn by turn.

**Kiffer:** Take care, everyone.

**Sarah:** See you in Lesson ten.