Labov and Waletzky Narrative Analysis: The Six Parts

Ask someone how their day went and you get a chronicle: this happened, then that. Ask about the time they nearly drowned and you get a story—one with a point, built to answer the unspoken challenge every storyteller faces: so what? William Labov and Joshua Waletzky showed that these oral stories of personal experience share a recurring architecture, and that architecture is codeable. Labov and Waletzky narrative analysis is the scheme for tagging that structure, and its lasting insight is that the part which turns events into a story is not the plot but the evaluation (Labov & Waletzky, 1967; Labov, 1972).

What is Labov and Waletzky narrative analysis?

Labov and Waletzky narrative analysis is a scheme that breaks a spoken personal-experience story into six functional parts, showing how a narrator turns raw events into a told story (Labov & Waletzky, 1967). The model was built inductively: Labov elicited stories in sociolinguistic interviews—famously with the “danger of death” question, “Were you ever in a situation where you thought you were in serious danger?”—and analyzed their recurring shape.

Two dating points keep the citation honest. The framework originates in Labov and Waletzky’s 1967 paper, but the canonical six-part structure and the full theory of evaluation are developed in Labov’s 1972 chapter “The transformation of experience in narrative syntax.” Cite the model as Labov and Waletzky 1967, elaborated in Labov 1972, rather than pinning all six parts to the earlier paper. The model was also built on one genre—oral, first-person narratives—so it does not transfer cleanly to fictional, written, or multi-party storytelling.

The six parts of a narrative

Each part answers a different question about how the story is built. The table defines each with the question it answers and a clause from one running synthetic narrative.

Part	Question it answers	Example clause
Abstract	What is this story about?	"So this one time I nearly drowned."
Orientation	Who, when, where?	"I was maybe ten, at my cousin's lake house, in August."
Complicating action	And then what happened?	"I swam past the dock, got a cramp, and went under."
Evaluation	Why does this matter?	"I honestly thought that was it."
Resolution	How did it turn out?	"My uncle pulled me out before I'd swallowed half the lake."
Coda	How do we get back to now?	"I still won't swim where I can't see the bottom."

Notice that a minimal narrative can consist of complicating-action clauses alone—“I swam out, got a cramp, went under, got pulled out.” It would be a valid sequence of events. What it would lack is any sense of why it was worth telling, which is the job of the next part.

Why is evaluation the heart of the model?

Evaluation is what distinguishes a narrative from a bare report, which is why Labov treated it as the center of the whole model (Labov, 1972). A chronicle lists what happened; a story tells you why it mattered. Evaluation is the set of devices—intensifiers, comparisons, asides, reported thought—through which the narrator signals significance and stance, defending the story against a listener’s “So what?”.

The complication for anyone coding it is that evaluation is rarely a tidy section sitting between the action and the resolution. Labov catalogued a graded range from external evaluation, where the narrator steps outside the story to comment (“it was the scariest thing that ever happened to me”), to fully embedded evaluation, folded into the action itself through the words chosen and the way events are intensified. Because it is distributed through the narrative rather than localized, evaluation is the hardest part to segment cleanly—and, as the reliability data below show, the hardest to code.

Complicating action vs evaluation: the coding discriminator

The discriminator between the two most important parts is temporal juncture. A complicating-action clause is a narrative clause: it is fixed in the event sequence, so that reversing two such clauses changes the listener’s understanding of what happened first (Labov, 1972). “I got a cramp and went under” means something different from “I went under and got a cramp.”

An evaluative clause is, by contrast, largely free. It suspends the forward motion of the story to comment on it, and you can move it or delete it without altering the order of events. “I honestly thought that was it” can sit almost anywhere in the drowning story without changing the sequence.

The practical coding rule follows directly: ask whether the clause advances the plot or comments on it. If reordering it would confuse the timeline, it is complicating action; if not, it is probably evaluation.

How reliably can narrative structure be coded?

Reliability depends heavily on which part you are coding: concrete, referential elements are agreed on easily, while interpretive ones are not. The sharpest illustration comes from a study comparing undergraduate scorers with an expert on narrative macrostructure elements—story-grammar categories rather than Labov’s own six parts, but the pattern carries across (Jones et al., 2019).

Coders matched the expert at a quadratic weighted kappa of 0.956 for Character but only around 0.40 for Plan and Consequence, the elements carrying goals and internal states (Jones et al., 2019). That gap is the empirical echo of Labov’s theoretical point: the evaluative, significance-bearing content is exactly what resists reliable coding. Purpose-built coherence schemes do better on their dimensions—the Narrative Coherence Coding Scheme reported intraclass correlations of roughly 0.80 for context, 0.82 for chronology, and 0.89 for theme (Reese et al., 2011)—and a narrative macrostructure composite reached Krippendorff’s alpha of 0.79, just under the conventional adequacy threshold (Heilmann et al., 2010). If you are computing those coefficients yourself, the Cohen’s kappa and inter-rater reliability guide covers why a low value on an interpretive category is a codebook signal, not a coder failure.

How is Labovian analysis used in speech-language pathology?

Narrative structure is a workhorse of speech-language assessment, because how a child or adult builds a story reveals language ability that isolated tests miss. Clinicians score narrative macrostructure—whether a told story has orientation, a complicating event sequence, and a resolution—as a window on discourse-level competence.

Two lineage notes matter for accuracy. Peterson and McCabe (1983) adapted Labov’s evaluation-centered approach into “high point analysis,” a developmental method organized around a narrative’s evaluative high point; it is a descendant of the model, not the original. And modern rubrics like the Narrative Scoring Scheme and the Monitoring Indicators of Scholarly Language (MISL) blend Labovian evaluation with Stein and Glenn’s story grammar (setting, initiating event, plan, consequence), so their categories map only loosely onto Labov’s six parts, and “evaluation” does not survive as a clean, isolable label (Heilmann et al., 2010). If you use these tools, know which tradition each category comes from.

How do you annotate narrative structure in a transcript?

To annotate a narrative, tag each clause or span with the part it performs—instance-mode annotation, where the label names the narrative function of the span rather than its content. Evaluation is the part to watch: code it wherever the narrator steps out of pure event-reporting to signal significance, even when that happens mid-action.

Here is the running story, fully labeled:

So this one time I nearly drowned. [abstract] I was maybe ten, at my cousin’s lake house, in August. [orientation] I swam out past the dock, got a cramp, and went under. [complicating action]

I honestly thought that was it. [evaluation] My uncle pulled me out before I’d swallowed half the lake. [resolution] I still won’t swim where I can’t see the bottom. [coda]

Tagging each span with its narrative function keeps the evidence attached to the label, so a second coder can check each call and compute agreement. That span-level discipline is what makes the Labov narrative coding scheme reproducible, and it is the same instance-mode logic behind labeling what an utterance does with Searle’s speech acts or the parts of an argument in the Toulmin model. Working with real interview or clinical narratives means handling sensitive speech, so the sane default is de-identified text; Tagaroo supports a browser-side anonymous mode so transcripts can stay local. The broader taxonomy of narrative and discourse schemes lives in the Tagaroo scale library.

Common mistakes when coding narrative structure

The recurring errors come from treating a fluid oral form as a rigid template:

• Expecting evaluation in one place. It is woven through the story, not parked between action and resolution. Code it wherever significance is signaled (Labov, 1972).
• Forcing an abstract or coda. Both are optional. A story without them is still a complete narrative.
• Confusing complicating action with evaluation. Use the temporal-juncture test: if reordering the clause would change the timeline, it is action, not evaluation.
• Equating the model with story grammar. Labov’s evaluation-centered scheme and Stein-Glenn story grammar are different traditions; modern SLP rubrics mix them (Heilmann et al., 2010).

The practical upshot: Labov and Waletzky narrative analysis endures because it found the part of a story that makes it a story. Code the six parts by the function each clause performs, lean on the temporal-juncture test to separate action from evaluation, and expect the evaluative core—the reason the story is being told—to be both the most important element and the hardest one to pin down.

References

• Labov, W., & Waletzky, J. (1967). Narrative analysis: oral versions of personal experience. In J. Helm (Ed.), Essays on the Verbal and Visual Arts (pp. 12–44). University of Washington Press. (Reprinted 1997, Journal of Narrative and Life History, 7(1–4), 3–38, doi:10.1075/jnlh.7.02nar.)
• Labov, W. (1972). The transformation of experience in narrative syntax. In Language in the Inner City: Studies in the Black English Vernacular (pp. 354–396). University of Pennsylvania Press.
• Labov, W. (1997). Some further steps in narrative analysis. Journal of Narrative and Life History, 7(1–4), 395–415. doi:10.1075/jnlh.7.49som
• Peterson, C., & McCabe, A. (1983). Developmental Psycholinguistics: Three Ways of Looking at a Child’s Narrative. Plenum Press. doi:10.1007/978-1-4757-0608-6
• Heilmann, J., Miller, J. F., Nockerts, A., & Dunaway, C. (2010). Properties of the Narrative Scoring Scheme using narrative retells in young school-age children. American Journal of Speech-Language Pathology, 19(2), 154–166. doi:10.1044/1058-0360(2009/08-0024)
• Reese, E., Haden, C. A., Baker-Ward, L., Bauer, P., Fivush, R., & Ornstein, P. A. (2011). Coherence of personal narratives across the lifespan: a multidimensional model and coding method. Journal of Cognition and Development, 12(4), 424–462. doi:10.1080/15248372.2011.587854
• Jones, S., Fox, C., Gillam, S., & Gillam, R. B. (2019). An exploration of automated narrative analysis via machine learning. PLOS ONE, 14(10), e0224634. doi:10.1371/journal.pone.0224634

If you code narrative structure in interviews or clinical language samples, Tagaroo turns the Labov narrative scheme into a guided, evidence-anchored annotation workflow—with inter-rater reliability computed as your coders work.