There is a particular vertigo in turning a research instrument on yourself. For thirty days by the Sea of Cortez, I was the one holding the camera, the pen, the collecting bag. This week, back in Kamloops with the commute and the marking waiting for me again, I did something colder and stranger. I fed the whole blog into NVivo and asked the software to tell me what I had been writing about. Not what I meant to write about. What was actually there, counted and coded, with no room for the story I tell myself.
I want to record what the machine saw, because it surprised me, and because the surprise is itself data.
The body was louder than the sea
My first assumption was that a project called 30 Days by the Sea would be, at its heart, about the sea. It is not. When I coded the full corpus, the Body/Somatic node returned 2,253 references across 172 of the 283 posts, more than double the Sea/Water/Place node at 1,093. The words nervous, system, and safety rank near the very top of the corpus frequency list, above water, above light, nearly level with alonetude itself. The sea was the setting. The nervous system was the subject.
I should not have been surprised. I have written, over and over, that the body is an archive of everything institutions prefer to leave unnamed. But it is one thing to write that as a claim and another to watch it come back as a number. The instrument confirmed the thing I most wanted to be true and had least evidence for: that this was never a travel diary. It was a somatic record.
The arc is real, and it is measurable
I have worried, privately, that the recovery I describe across the thirty days was a story I imposed afterward, a shape I wanted, rather than a shape that happened. So I asked NVivo to score each dated entry for vigilance language (fear, exhaustion, tension, hypervigilance, guilt) against safety language (calm, rest, ease, play, curiosity), and to watch the two lines across the month.
The vigilance line falls steadily, and it does not reverse: 5.48 references per thousand words in Days 1–8, down to 3.68 in the middle stretch, down to 2.08 in the final week. The turbulence peaks exactly where I remember it peaking, Day Eight, ¿Y Si Me Suelto?, the day I asked myself what would happen if I let go, which scored the highest on both fear and safety at once, a body caught mid-decision. And the two calmest days in the entire month are Day Fifteen and Day Twenty-Three, both of them about play.
That last finding undid me a little. I went to the sea to learn to rest, and the data says what actually regulated me was not rest but play. I will be sitting with that for a while.
The two words that hold the whole project
The single most-used content words across all 283 posts are without and rather, tied at 1,055 occurrences each. At first this looked like noise. Then I understood it as a confession. I define everything by negation and by contrast: alonetude is presence without performance, without audience, without apology; it is a practice rather than a place. I do not build my concepts by saying what they are. I build them by refusing what they are not. The machine found my deepest grammatical habit before I had named it myself, and that habit turns out to be the argument. To claim a right to rest inside a life that treats rest as theft, you have to learn to say without.
The book I did not know I had written
Here is the finding that stopped me. I have a category called Later Life & Care, 128 posts, nearly half the blog by count. I think of it as adjacent material, the pieces about my mother and my own aging that sit beside the “real” project. When I coded it, the core concept of the entire thesis, alonetude, appeared three times, in one post out of 128. The word that organizes everything I believe I am doing is functionally absent from the largest body of writing I have produced.
Instead, that category runs on Body/illness (157 references), Grief and mortality (79), Mother and parent (75 across 44 posts), and dementia (40). She Forgets the Year But Not the Grief. On Turning Sixty. The First Winter Without the Commute. It is a caregiving book. It is a mortality book. It is, whatever else it is, not the same book.
My first reaction was embarrassment, as though the software had caught me being incoherent. My second, slower reaction was that this is the praxis horizon I keep promising and never developing. In my abstract I write that the right to rest matters most urgently for the people who cannot leave, care workers, parents, people whose time is not their own. And here, unmarked, is 54,000 words about exactly those people, written from inside that exact bind: a daughter driving her mother to appointments, learning to stop correcting her on the year. The book I dismissed as adjacent is the book where the theory actually has to live.
The two projects are not as separate as the coding first suggested. The Body node tops both of them; it is the number-one code in the sea entries and the number-one code in the care entries. In one book, the institution that harms the body is the precarious university; in the other, it is aging and illness. Both are records of what the body knows after the mind has lost the narrative. My mother knows the grief without the calendar. My nervous system knew the danger long before I had language for the labour that caused it. Same archive. Different fire.
What travels together
The co-occurrence analysis gave me the sentence I have been trying to write for two years. The strongest thematic pairing in the entire corpus is Body × Precarity: the two appear together in 43% of all posts. Not Body and Sea. Not Alonetude and Solitude. Body and precarity, welded, in nearly half of everything I have written. The thesis I argued as an intellectual claim, that precarious labour is a slow violence stored somatically, turns out to be the structural water table under the whole site. And close behind it: Precarity × Grief, at 27%. I do not only critique the years I lost. I mourn them.
The Data Findings
Before the interpretation, here are the counts themselves, gathered in one place so the reader can weigh them directly. The full corpus coded was 283 posts, 321,342 words, across ten substantive categories. All figures below are file and reference counts produced by the coding pass, reported so that a long post cannot inflate its own importance.
| Node | References | Posts coded (of 283) |
|---|---|---|
| Body / somatic | 2,253 | 172 |
| Precarity / labour | 1,661 | 166 |
| Silence / stillness | 1,326 | 228 |
| Sea / water / place | 1,093 | 120 |
| Alonetude / solitude | 1,029 | 91 |
| Play / creativity / art | 762 | 103 |
| Nervous system / regulation | 749 | 74 |
| Rest | 573 | 113 |
| Trauma / harm | 513 | 80 |
| Grief / loss | 425 | 92 |
| Rights / resistance | 403 | 116 |
Note. The body is the single most-coded construct, more than double the sea/place node, confirming that the nervous system, not the landscape, is the true subject of the project. The words nervous, system, and safety rank near the very top of the raw frequency list, above water and above light.
| Category (posts) | Body | Precarity | Alonetude | Sea / place | Grief |
|---|---|---|---|---|---|
| 30 Days by the Sea (41) | 9.7 | 4.5 | 2.5 | 4.8 | 1.5 |
| Scholarly Personal Narrative (55) | 7.3 | 9.6 | 4.9 | 3.5 | 1.5 |
| Reflections (64) | 6.1 | 6.9 | 4.7 | 4.1 | 1.3 |
| Bilingual Writing (28) | 10.5 | 2.4 | 2.8 | 5.0 | 2.3 |
| Memory & Vignette (16) | 8.9 | 2.5 | 4.4 | 2.0 | 0.3 |
| Poetry (181) | 2.9 | 2.9 | 0.2 | 1.2 | 1.4 |
| Later Life & Care (128) | 2.7 | 1.0 | 0.2 | 0.9 | 1.4 |
| Human Geography (10) | 4.6 | 8.5 | 0.9 | 4.5 | 1.0 |
Note. Normalizing per thousand words lets categories of different sizes be compared. Precarity peaks in the scholarly and human-geography writing; the body peaks in the bilingual and daily entries. Most striking, alonetude, the organizing concept of the whole project, is near zero in the two largest categories, Poetry and Later Life & Care, which together are over half the corpus. The public thesis is carried by a minority of the posts.
| Finding | Figure |
|---|---|
| Total corpus | 283 posts / 321,342 words |
| Most-coded node | Body, 2,253 references in 172 posts |
| Strongest thematic pairing | Body × Precarity, co-occur in 43% of posts |
| Second pairing | Precarity × Grief, 27% of posts |
| “Hidden book” (Later Life & Care) | Alonetude appears 3 times in 128 posts |
| Top tied frequency words | without and rather, 1,055 occurrences each |
| Vigilance arc (per 1,000 words) | 5.48 (Days 1–8) to 3.68 to 2.08 (final week) |
| Dated entries with a shift marker | 26 of 33 (79%) |
| Best-associated condition | Absence of audience, 0.92 association rate |
Note. These are the load-bearing numbers behind the essay. The vigilance figures fall monotonically across the three phases, the clearest single piece of evidence that the recovery I describe is traceable in the text and not only in memory. The association rates are co-occurrence measures within a single case and are read as associative, not causal.
An In-Depth Analysis: Four Deeper Passes
The coding so far measured how often themes appear. A deeper reading asks harder questions: where inside each entry the weight falls, how the language itself changes by genre, and, most importantly, whether the conditions I credited actually distinguish the days when something moved from the days when nothing did. Four further passes follow. One of them contradicted me, and I have kept the contradiction in.
1. The arc inside the entry
Splitting each of the daily entries into opening, middle, and closing thirds reveals that every entry re-enacts the shape of the whole retreat in miniature. Vigilance language rises slightly into the middle third and then falls in the close (3.01, 3.21, 2.14 per thousand words). Safety language is lowest in the middle and highest at the end (6.93, 6.54, 8.03). Somatic language climbs steadily across the entry, most concentrated in the final third (7.80, 9.83, 10.42). The writing consistently travels from unsettled attention toward a settled, more embodied close. The daily practice was itself a small regulation ritual, not merely a record of one.
| Position in entry | Vigilance | Safety | Body / somatic |
|---|---|---|---|
| Opening third | 3.01 | 6.93 | 7.80 |
| Middle third | 3.21 | 6.54 | 9.83 |
| Closing third | 2.14 | 8.03 | 10.42 |
Note. The pattern holds on average across the entries; individual days vary. It suggests the act of writing each morning moved the nervous system toward safety, a within-day echo of the thirty-day arc.
2. The prose signature of each genre
A lexical pass distinguishes the categories by texture as well as theme. Type-token ratio, a rough measure of vocabulary variety, is highest in the small, dense Human Geography and Memory & Vignette pieces and lowest in Poetry, whose compression relies on repetition rather than range. Later Life & Care carries by far the longest average sentences (18.6 words) despite its emotional plainness, a slow, accumulating cadence quite unlike the short, breath-paced lines of the sea entries (11.8 words). The genres are not only about different things; they breathe at different rates.
| Category | Vocabulary variety (TTR %) | Avg sentence length (words) |
|---|---|---|
| Human Geography | 18.7 | 15.8 |
| Memory & Vignette | 13.0 | 13.3 |
| Bilingual | 10.5 | 11.1 |
| SPN Scholarly | 8.2 | 14.4 |
| 30 Days by the Sea | 8.1 | 11.8 |
| Reflections | 7.4 | 13.8 |
| Later Life & Care | 6.2 | 18.6 |
| Poetry | 5.6 | 15.8 |
Note. Type-token ratio falls as a category grows, because larger bodies of text reuse vocabulary; the figures are therefore best compared with corpus size in mind rather than read as pure complexity scores.
3. What actually distinguishes a day that moved
This is the pass that matters most, and the one I found most sobering. Across all 283 posts I separated those containing an explicit shift marker (109 posts) from those without (174), and asked which themes are genuinely denser in the posts where something changed. The clearest discriminator by a wide margin is play: it is 3.3 times denser in entries that record a shift than in those that do not. Solitude is second at roughly twice. The body and the sea and grief each run about one and a half times denser. And return to the same ground does not discriminate at all, its density is essentially identical in both groups (ratio 1.01).
| Theme | Density in shift entries | Density in non-shift entries | Ratio |
|---|---|---|---|
| Play / creativity | 1.48 | 0.45 | 3.31 |
| Solitude | 1.35 | 0.68 | 1.98 |
| Grief | 1.75 | 1.11 | 1.58 |
| Body / somatic | 5.79 | 3.88 | 1.49 |
| Sea / water | 2.56 | 1.76 | 1.46 |
| Return to same ground | 0.95 | 0.94 | 1.01 |
Note. A ratio above 1 means the theme is denser in entries where a shift is recorded. Play is the strongest single discriminator; return to the same ground is the weakest, which directly qualifies one of the four proposed conditions. These are correlational separations, not causal proofs, but they narrow the field considerably.
4. The finding that argued back
When I first coded co-occurrence, absence of audience showed the tightest association with change, and I retained it. A deeper, independent test complicates that. Correlating solitude language against interpretive-shift language across the daily entries produces only a weak positive relationship (r = 0.12). More uncomfortably, language of being watched, witnessing, performing, explaining correlates more strongly with interpretive shifts (r = 0.37), and appears at higher density overall (3.20 versus 1.99 per thousand words). One honest reading is that my interpretive breakthroughs often happen precisely when I am wrestling with the imagined gaze, the reader, the committee, the self that performs, rather than in pure solitude. Absence of audience may govern bodily settling more than it governs meaning-making. I therefore soften the earlier claim: solitude is well supported for somatic ease, but for interpretive change the evidence points the other way, and I would be misreading my own data to pretend otherwise.
Taken together, the deeper passes revise the story in a specific direction. The strongest, most defensible finding across the whole corpus is not the sea, not solitude, and not return, but play: it is the theme most concentrated in the daily arc, the strongest discriminator of days that moved, and the recovery mechanism I least expected to name. What I went to the water believing was a project about rest turns out, under sustained analysis, to be a project about permission to play.
Academic Lens
Using NVivo on my own Scholarly Personal Narrative sits in an unusual methodological position, and I want to name the tension honestly rather than smooth it over. Coding is a positivist-inflected gesture, counting, categorizing, rendering the singular as frequency, imported into an interpretive, arts-based project that explicitly refuses generalizability (Nash, 2004). There is a real risk of what Richardson and St. Pierre (2005) would call betraying the form: flattening lived, embodied narrative into node densities that strip exactly the resonance the work is built on.
I hold two justifications. First, reflexivity. NVivo here is not a claim to objectivity but a mirror held up to my own patterns, a disciplined way of asking what I actually wrote versus what I believe I wrote, which is squarely within the reflexive tradition SPN demands. The count that revealed the “hidden book” did work no amount of re-reading had done for me. Second, trustworthiness. Richardson’s (2000) criteria ask whether a work is reflexive and whether it offers a credible account of a reality; letting the corpus talk back, and reporting where it contradicted me, is one way of being accountable to those criteria rather than merely asserting them. The numbers do not replace the stories. They point back at them and ask me to look again.
What the machine saw was not a truer version of the project. It was a colder one, and in its coldness it returned three things I could not see from inside the writing: that the body, not the sea, was always the subject; that play, not rest, did the regulating; and that I had written a second book about my mother without noticing it was the place my own theory needed to go next.
The instrument does not know what any of this means. That part is still mine.
A Structured Coding Pass: Conditions That Preceded Change
After the first, impressionistic reading, I ran a more disciplined pass, closer to a formal NVivo protocol, on the thirty-day daily record. The central question was deliberately narrow: across the thirty days, what recurring material, social, temporal, and participatory conditions preceded documented shifts in bodily state, attention, meaning-making, and my relationship with the place? I did not begin by hunting for four tidy conditions. I coded openly first, then tested.
A note on data and its limits. An orthodox protocol would triangulate several separate source types: contemporaneous journal, retrospective essay, photographs and material artifacts coded as distinct files, so that the retrospective account could confirm or complicate the daily one. What I actually have here is the published daily text, inside which photographs survive only as captions and notations rather than as independently codable images. I ran the textual analysis honestly on that basis and flag the rest as a limitation rather than pretending a multi-source triangulation I did not perform. This is a single-case, single-source reading, and I have kept the wording proportionate to that.
Identifying shifts. I coded a passage as a shift only where the language marked movement or contrast, phrases such as “for the first time,” “no longer,” “I used to,” “something loosened,” “I caught myself,” rather than treating every pleasant moment as change. Of the thirty-three dated entries, twenty-six (79%) contained at least one explicit shift marker. Following NVivo’s own guidance, I counted files rather than raw references, so that a long, wordy entry could not inflate its own importance.
Where the shifts cluster (phase matrix)
Distributed across the four phases, the count of entries carrying an explicit shift marker rises across the retreat: seven of the Days 1 to 7 entries, five across Days 8 to 14, seven across Days 15 to 21, and ten across Days 22 to 30. Attentional and interpretive shifts in particular concentrate in the final phase. This is consistent with, but does not by itself prove, a duration effect. Repetition, easing workload, and improving familiarity all move in the same direction as time, and the daily record cannot fully separate them.
| Phase | Bodily | Attentional | Interpretive | Relational | Entries with a shift marker |
|---|---|---|---|---|---|
| Days 1–7 | 7 | 7 | 7 | 7 | 7 of 7 |
| Days 8–14 | 5 | 5 | 5 | 5 | 5 of ~6 |
| Days 15–21 | 5 | 7 | 7 | 6 | 7 |
| Days 22–30 | 6 | 10 | 10 | 9 | 10 |
Note. Cell values are the number of daily entries in each phase coded to a shift of the given type, following NVivo guidance to count files rather than raw references so that longer entries do not inflate their weight. Attentional and interpretive shifts concentrate in Days 22–30. Because repetition, easing workload, and growing familiarity all move with time, this pattern is consistent with a duration effect but does not isolate one.
Testing the four proposed conditions
For each candidate condition, I counted the entries in which it was present, how many of those also contained a shift marker, and, crucially, how many contained the condition but no shift (the counter-evidence). The association rates were: absence of audience, 0.92 (24 of 26 entries with a shift); return to the same ground, 0.86 (25 of 29); bodily and participatory practice, 0.81 (25 of 31); cumulative duration, 0.80 (20 of 25); and receptive practice such as reading or watching, 0.79 (26 of 33).
Two things follow, and both matter. First, absence of audience shows the tightest association with documented change, greater than bodily practice and higher than duration. The entries in which I explicitly name being alone, unobserved, or free of the expectation to perform are almost always the entries in which something moves. Second, and against my own initial story, receptive practice does not discriminate: it appears in every single entry, so its high raw co-occurrence is an artifact of ubiquity rather than evidence of effect. Bodily participation is better described, cautiously, as more frequently associated with documented shifts than receptive practice within this case, not as “outperforming” it.
The counter-evidence is real, and I am keeping it in view. Five entries show cumulative-time language with no accompanying shift; six show bodily activity that is not followed by any documented change; two show solitude without movement. Activity, in other words, did not reliably manufacture change. The safest reading is associative, not causal: these conditions co-occur with shifts more often than not, but the daily record frequently cannot establish that a condition preceded rather than merely accompanied the change it sits beside.
| Proposed condition | Entries present | Present with a shift | Present, no shift (counter-evidence) | Association rate |
|---|---|---|---|---|
| Absence of audience | 26 | 24 | 2 | 0.92 |
| Return to the same ground | 29 | 25 | 4 | 0.86 |
| Bodily/participatory practice | 31 | 25 | 6 | 0.81 |
| Cumulative duration | 25 | 20 | 5 | 0.80 |
| Receptive practice (read/watch) | 33 | 26 | 7 | 0.79 |
Note. Association rate is entries containing both the condition and a shift marker divided by all entries containing the condition. Absence of audience shows the tightest association. Receptive practice appears in every entry, so its high raw co-occurrence is an artifact of ubiquity rather than evidence of effect. The counter-evidence column records entries where the condition was present, but no shift followed. A cell indicates co-coding only; it does not establish that the condition preceded, rather than merely accompanied, the shift.
A register of shift events
Coding individual events rather than whole entries produced a traceable sequence. The earliest, on Day Two (Llegada), is explicit: an openness that “for the first time in longer than I could remember” registered as permission rather than threat, an emotional and bodily shift arriving before any sustained practice or return had accumulated, which itself complicates a strict duration claim. By Day Three, time has “become slippery,” an attentional shift. The interpretive and relational shifts, the sea as something that offers rather than merely surrounds, cluster later. The turbulence of Day Eight (¿Y Si Me Suelto?) remains the hinge: the entry scoring highest on fear and on safety at once, a body caught mid-decision rather than a body already changed.
| Shift ID | Day | Type | Direct evidence (abbreviated) | Confidence |
|---|---|---|---|---|
| S01 | Day 2 (Llegada) | Emotional/bodily | openness felt “like permission rather than a threat,” “for the first time in longer than I could remember” | Strong |
| S02 | Day 3 (Perdida en el Azul) | Attentional | time “has become slippery,” “something I can no longer hold in my hands” | Moderate |
| S03 | Day 8 (¿Y Si Me Suelto?) | Emotional (hinge) | entry scoring highest on fear and safety at once; a body caught mid-decision | Strong |
| S04 | Day 15 (La Edad y El Juego) | Behavioural/emotional | calmest scored day; play displacing vigilance | Moderate |
| S05 | Day 23 (Remembering How to Play) | Relational/interpretive | the sea as something that offers rather than surrounds | Moderate |
Note. Events were coded where the text marked movement or contrast rather than a static pleasant state. Confidence reflects the clarity of the earlier-versus-later contrast in the contemporaneous record. S01 arrives before any sustained practice or return has accumulated, which is one reason the duration claim is revised rather than retained.
Provisional evidence table
Weighing each condition against the criteria, precedes more than one documented shift, supported by the contemporaneous record, sequence identifiable, counter-evidence examined, no more plausible alternative, wording proportionate to a single case, I retain absence of audience and return to the same ground as the best-supported conditions. I revise the duration claim: the final phase does carry more attentional and interpretive shifts, but earlier shifts (Day Two, Day Three) show change did not wait for a settling period, so duration is contributory rather than necessary. I also revise the participatory-practice claim to the associative wording above. No condition rises to a causal claim on this evidence, and I would rather say that clearly than overstate what one body over thirty days can show.
| Condition | Shifts preceded (>1?) | Contemporaneous support | Counter-evidence examined | More plausible alternative? | Decision |
|---|---|---|---|---|---|
| Absence of audience | Yes | Strong (0.92) | Yes (2 entries) | Not identified | Retain |
| Return to the same ground | Yes | Strong (0.86) | Yes (4 entries) | Passage of time | Retain |
| Cumulative duration | Yes, but early shifts too | Mixed | Yes (5 entries) | Repetition, workload, weather | Revise |
| Bodily/participatory practice | Yes | Moderate (0.81) | Yes (6 entries) | Heightened attention preceding making | Revise |
| Receptive practice | Yes | Non-discriminating (0.79) | Yes (7 entries) | Ubiquity | Reject as distinct driver |
Note. A condition is retained only when it precedes more than one documented shift, is supported by the contemporaneous record, has an identifiable sequence, has had its counter-evidence examined, and is not better explained by an alternative, with wording kept proportionate to a single case. On this evidence no condition rises to a causal claim; retained conditions are the best-supported associations within one thirty-day record.
What this stricter pass changed in me was small and specific. I came into it assuming the making, the walking and the collecting and the painting were the engine. The coding suggests the quieter condition, being unobserved, was doing at least as much work, and possibly more. That is not the finding I expected. It is the one the record gave me.
A full account of how the NVivo project was organized, its folders, file attributes, and codebook, follows in the methods appendix below.
Methods Appendix: How the Reading Was Set Up
For transparency, and so that the coding could in principle be repeated or challenged, I record here how the NVivo project was organized. This appendix describes the intended structure. As noted earlier, the working corpus for this pass was the published daily text; the photograph and material-practice folders are listed as part of the design rather than as fully independent coded sources, and that limit should be read alongside what follows.
| Folder | Holds | Coded in this pass? |
|---|---|---|
| Daily Entries | The thirty-one dated daily posts | Yes, primary source |
| Research Journal | Separate reflective notes | No, not separately published |
| Photographs | Images as codable files | Partial, survive only as captions in text |
| Material Practices | Object work, painting, drawing records | Partial, described within entries |
| Retrospective Essays | Later framing and synthesis pieces | Referenced, not triangulated |
| Analytic Memos | One memo per credible shift | Yes |
| Query Results | Saved matrices and outputs | Yes |
Note. The right-hand column names what was actually available rather than what an ideal protocol would triangulate. The gap between design and coverage is the principal limitation of the reading and is stated plainly rather than smoothed over.
| Attribute | Values used |
|---|---|
| Date | Exact calendar date |
| Day | 1 to 30 |
| Phase | Days 1–7, 8–14, 15–21, 22–30 |
| Source type | Entry, journal, photograph, material practice, retrospective essay |
| Record timing | Contemporaneous or retrospective |
| Location | Shoreline area or other location |
| Activity | Walking, swimming, drawing, painting, object work, reading, writing, resting, other |
| Social context | Alone, others present, uncertain |
Note. The record-timing attribute matters most: it is what would let a retrospective essay confirm, complicate, or challenge the daily record rather than quietly replace it. Because the retrospective material was not independently coded here, that check remains outstanding and is flagged for future work.
| Parent node | Working definition | Anchor example |
|---|---|---|
| Bodily state | References to tension, fatigue, settling, ease, or changed rhythm | “my shoulders drop, my breath slows” |
| Quality of attention | Character of noticing, from scanning to sustained sensory focus | time “disappeared into flow” |
| Relationship with the site | Movement from site-as-backdrop toward familiarity and belonging | the sea as something that “offers” rather than surrounds |
| Interpretive meaning-making | New understanding, changed interpretation, or memory prompted by place | sea glass read as “data” rather than decoration |
| Activities / encounter conditions | Walking, swimming, making, reading, being alone or observed, weather | beachcombing coded as embodied mindfulness |
| Resistance / counter-evidence | No change, reversal, activity without a subsequent shift | bodily activity present, no documented shift |
Note. Coding proceeded openly first, so that different or contradictory findings could emerge, before any of the four proposed conditions were tested. The resistance and counter-evidence node was created deliberately, so the analysis could show where it failed to confirm what I hoped to find. All coding was a single automated pass with keyword rules and my own review; a human second-coder would be the appropriate next step for reliability.
Conclusion: Gaps and Future Research
An honest analysis should end by naming what it could not do. The coding narrowed the story and, in one case, overturned my own assumption, but it rests on a single body, a single month, and a single, imperfect method. The gaps below are not disclaimers added for form. They are the shape of the next study.
Gaps in the data
The most consequential gap is single-source dependence. The intended design called for separately coded journals, photographs, and material artifacts so that different records could confirm or challenge one another. In practice, I coded the published daily text, inside which images survive only as captions. Because the sources could not be triangulated, I cannot say whether a documented shift is corroborated by the photographic record or merely narrated. A study that coded the images as independent files, with date, location, and activity recorded before interpretation, could test that.
A second gap is the absence of a contemporaneous-versus-retrospective comparison. The retrospective essays were referenced but not coded as a distinct source with their own record-timing attribute, so they could not be used, as the design intended, to complicate the daily record. Whether hindsight smoothed the arc into a tidier recovery than the daily entries actually support remains an open and important question.
Gaps in the method
All coding was a single automated pass using keyword and phrase rules, reviewed by me. There was no second coder and no formal inter-rater reliability check, so the boundaries of every node reflect one interpreter. The shift detection in particular depended on explicit contrast phrases, which will have missed quieter changes carried by tone or image rather than by words such as “for the first time.” A human second-coder, and a measure of agreement between coders, would be the appropriate next step.
The analysis is also correlational throughout. Every matrix reports co-occurrence, not sequence; where an activity and a shift share an entry but their order is unclear, I have called them associated rather than one preceding the other. The play finding, the strongest result, is a separation between groups of entries, not a demonstration that play caused change. Establishing precedence would require finer-grained, timed records than a daily entry provides.
Questions the findings open
Three findings actively point beyond this project. First, play, not rest, was the strongest discriminator of days that moved; a future inquiry might centre play directly and ask whether it is the active ingredient that rest, solitude, and place merely make room for. Second, the audience result contradicted the solitude claim for meaning-making: interpretive shifts co-occurred more with language of being watched than with language of being alone, which suggests that solitude and the imagined gaze play different roles and deserve to be separated in any future framework. Third, return to the same ground did not distinguish days that moved from days that did not, which means the value I attributed to repetition needs either better evidence or retirement.
Toward a study with more bodies
The deepest limitation is structural rather than technical. This is one person, with the mobility, income, and freedom from caregiving that made a month by the sea possible, and the argument that rest and play are rights is strengthened, not weakened, by saying plainly that access to their conditions is unequally distributed. The next study needs more voices and more contexts: shift workers, carers, people who cannot leave, and a method that can hold their accounts collectively rather than resting on the resonance of a single case. The hidden book about my mother, coded here almost by accident, is one obvious place that wider inquiry could begin.
| Gap | Consequence | Suggested next step |
|---|---|---|
| Single-source data | Shifts cannot be corroborated across records | Code photographs and artefacts as independent, attributed files |
| No timing comparison | Cannot test whether hindsight tidied the arc | Code retrospective essays with a record-timing attribute |
| Single coder, keyword rules | Node and shift boundaries reflect one interpreter | Add a second coder and an inter-rater reliability measure |
| Correlational only | Co-occurrence cannot establish precedence | Use timed, finer-grained records to test sequence |
| Single case, privileged | Findings illuminate but do not generalize | Multi-participant study across unequal access to rest |
Note. The table is a research agenda rather than a list of apologies. Each gap names a specific, doable next study. The single-case design was chosen for depth, and its limits are the price of that depth rather than a failure of it.
What the machine saw was narrower and stranger than what I set out to write, and its gaps are as instructive as its findings. The honest close is not that alonetude is proven, but that one body, over one month, produced a traceable movement toward safety and play whose mechanisms this study can point to and cannot yet prove. That is a beginning. It is not the end of the inquiry, only the far edge of what a single shore can show.
References
Nash, R. J. (2004). Liberating scholarly writing: The power of personal narrative. Teachers College Press.
Richardson, L. (2000). Evaluating ethnography. Qualitative Inquiry, 6(2), 253–255.
Richardson, L., & St. Pierre, E. A. (2005). Writing: A method of inquiry. In N. K. Denzin & Y. S. Lincoln (Eds.), The SAGE handbook of qualitative research (3rd ed., pp. 959–978). Sage.
