It happens every week in thousands of pediatric clinics. A parent—call her Miriam—sits in a plastic chair while her two-year-old, Leo, stacks blocks on the exam table. The nurse hands her a developmental screening form. Miriam reads the item: “Speaks in 3-word sentences by 24 months.” She glances at Leo, who just said, “Mira, mira, agua gone,” mixing Spanish and English the way he always does at home. Three words? Technically, yes. But not in the way the form imagines. Miriam freezes. She doesn’t know whether to check the box or leave it blank. The label that will follow—delayed, on track, needs evaluation—hangs in the air before she’s even made a mark.
I’ve watched versions of this scene unfold hundreds of times, first as a community clinic pediatrician and now as a researcher who studies why the tools we use to measure children so often miss the children themselves. The checklist is not neutral. It was built somewhere, by someone, with a particular child in mind. And the act of naming a behavior—calling it “delayed” or “advanced” or “age-appropriate”—is a small but powerful intervention. A label can open a door to early support, or it can close one by framing a child as a problem before anyone has asked what the child is actually doing.
This article is about that act of naming. It’s about how the labels we assign to childhood behaviors are products of the tools we use to name them, and how those tools have a history, a set of assumptions, and a quiet authority that can pre-write a child’s story. And because I’ve learned that the best way to see a familiar problem clearly is to hold it next to something unexpected, I want to draw a parallel to a world that might seem far from pediatric clinics: the world of writing and creativity, where a working title for a book or project can similarly shape—and sometimes distort—what follows.
The Checklist’s Hidden Biography
Most parents encounter developmental milestones as a list that arrives from the pediatrician’s office, a parenting app, or a well-meaning relative. The list feels timeless, as if it were handed down by nature. But milestone checklists have a specific biography. They trace back, in large part, to the work of Arnold Gesell, a psychologist and pediatrician who, in the 1920s, set up a dome-shaped observation laboratory at Yale. Gesell and his team filmed hundreds of children—mostly white, mostly from educated, two-parent families in New Haven—and catalogued the ages at which they rolled over, sat up, walked, and spoke. His methods were meticulous for their time, but his sample was narrow. The norms he published became the scaffolding for decades of developmental screening tools, even as the population of children grew vastly more diverse.
Today’s checklists have been updated, but the updates often remain tethered to the same logic: collect data from a reference population, calculate averages, and then treat deviation from those averages as a signal of possible delay. The problem is that the reference population still tends to be monolingual, Western, and drawn from households with resources that allow participation in research studies. A 2018 review in Pediatrics found that many commonly used developmental screening tools were normed on samples that underrepresented children from low-income families, bilingual homes, and racial and ethnic minority groups. The result is a tool that can produce “abnormal” results for children who are simply living in different contexts.
Leo, the two-year-old mixing Spanish and English, is a case in point. Bilingual children often distribute their vocabulary across two languages. They might know “dog” in English and “perro” in Spanish, but a checklist that counts words in only one language will underestimate their total lexicon. Research on bilingual language development consistently shows that when you measure what a child knows across both languages, the apparent “delay” often disappears. But the checklist doesn’t ask. It just offers a box.
When a Label Pre-Writes the Story
Labels have momentum. Once a child is described as “speech-delayed” in a medical record, that phrase travels. It appears in referral forms, in insurance pre-authorizations, in the notes of a new provider who hasn’t yet met the child. It shapes what people look for and what they stop looking for. A teacher who reads “delayed” may interpret a child’s quietness in class as evidence of a language problem rather than as a sign of a child who is listening, processing, or simply new to the social rules of a classroom. A parent who hears “delayed” may begin to interact with the child differently—offering fewer open-ended questions, waiting less time for a response—in ways that inadvertently narrow the child’s opportunities to practice language.
This is not an argument against early identification. Catching real delays early can change a child’s trajectory. But it is an argument for understanding that the act of naming is itself an intervention, one that carries risks alongside benefits. The label can become the story before the child has had a chance to tell their own.
I’ve seen the opposite happen, too. A child whose behavior is labeled “age-appropriate” in a brief screening may be quietly struggling in ways the checklist wasn’t designed to detect. A child who speaks in full sentences at the expected age but cannot take turns in conversation, or who uses scripted phrases from television rather than generative language, may pass the milestone item while missing something more fundamental. The label “on track” can close inquiry just as effectively as “delayed” can.
The Parallel That Surprised Me
Some years ago, while working on a writing project of my own, I found myself stuck on a title. I had a draft I believed in, but every title I tried felt either too vague or too narrow. A friend who works in publishing told me something I’ve never forgotten: “A working title is not the book. It’s a promise you’re making to yourself about what the book might become. But if you cling to it too early, you’ll write toward the title instead of toward the truth.”
I began to see the connection. A developmental label—delayed, advanced, age-appropriate—functions much like a working title. It is a provisional name, a placeholder that summarizes what we think we’re seeing. But once it’s written down, it starts to organize the narrative. It pre-writes expectations. It can foreclose possibilities that the actual story—the actual child—might otherwise have revealed.
This parallel is not just a metaphor. In both domains, the tools we use to generate names shape what we notice and what we overlook. Consider the world of book title generators. When a writer uses a tool like Reedsy’s Book Title Generator, the tool asks for genre, core conflict, and tone, then returns ten options, each with a one-line explanation of what it captures. The generator’s creators are careful to note that it is “better at sparking a direction than landing the final answer.” They point out that F. Scott Fitzgerald cycled through a dozen working titles—including Gold-Hatted Gatsby and Trimalchio in West Egg—before arriving at The Great Gatsby. The title was provisional, revisable, and never the whole story.
Developmental checklists, by contrast, are rarely presented as provisional. They arrive with the authority of medicine, printed on official forms, scored with cutoffs that trigger referrals. But they, too, are generators: they generate labels from the inputs they’re given, and those labels are only as good as the reference population and the assumptions built into the tool. A checklist normed on monolingual English-speaking children will generate “delayed” for a bilingual child not because the child is delayed, but because the generator was never calibrated for that input.
What the Tools Assume
Let’s get more concrete. Many developmental screening tools include an item like “points to at least one body part when asked.” The item assumes a particular kind of parent-child interaction: one in which an adult asks a direct question and waits for a child to point. In some families, children learn body parts through songs, through dressing routines, through bathing—not through quizzing. A child who knows where their ears are but has never been asked “Where are your ears?” in a testing situation may fail the item. The tool measures not just the child’s knowledge but the child’s familiarity with the testing format.
Similarly, items about social interaction often assume a two-parent household or a particular family structure. “Imitates household activities, like sweeping or feeding a doll” presumes the child has observed those activities and has access to the props. A child living in a shelter, a child whose primary caregiver works long hours, a child being raised by a grandparent with limited mobility—these children may show their developmental competence in ways the checklist never thought to ask about.
The problem is not that checklists exist. The problem is that we forget they are tools with histories and biases, and we treat their outputs as facts rather than as generated suggestions. This is where the parallel to AI tools becomes instructive. The Authors Guild, in its AI Best Practices for Authors, warns that AI outputs are “generic mashups of pre-existing works ingested during training” and that when a writer relies too heavily on such tools, they risk losing their “unique voice, thinking, and creativity.” The Guild’s guidance is not a blanket rejection of AI but a call to treat its outputs as provisional, to preserve human judgment, and to remember that the tool was trained on a corpus that may not represent the writer’s own context or community.
Developmental screening tools are not AI, but they share a structural similarity: they were “trained” on a particular dataset, and they generate outputs that can look objective while carrying the biases of their training data. When we treat those outputs as final, we make the same mistake a writer makes when they accept a generated title without asking whether it fits the book they’re actually writing.
Small Interventions That Change the Naming Process
So what can a parent, teacher, or clinician do? The answer is not to abandon screening tools—they catch real problems that might otherwise be missed. The answer is to change how we hold them.
One small intervention I’ve seen work in clinics is what I call the “translation pause.” Before a clinician discusses a screening result with a family, they take thirty seconds to ask: “What does this item actually measure in your home? Is there a different way your child shows this skill?” For Miriam and Leo, that pause might reveal that Leo speaks in three-word combinations all the time—just not in a single language. The label shifts from “delayed” to “bilingual, needs a different measure.”
Another intervention is to treat the screening form as a conversation starter rather than a verdict. Some pediatricians I’ve worked with have begun handing the checklist to parents with a simple script: “This is a tool that helps us notice things. It’s not a test your child passes or fails. Let’s go through it together and you tell me where it fits and where it doesn’t.” That reframing changes the power of the label. It becomes a working title, not a final diagnosis.
For teachers, a related practice is to write observational notes about a child before looking at any screening data or previous labels. What do you actually see? What does the child do when they’re comfortable, when they’re with peers, when they’re outdoors? Write that down first. Then read the label. The gap between the two is often where the most important information lives.
For parents, the most practical shift is to ask one question when a label is offered: “What did you observe that led you to that word?” A label like “delayed” is a summary. The observations behind it—the specific behaviors, the context, the comparison group—are the data. Asking for the data turns a label back into a conversation.
The Working Title Mindset
I’ve started to think of this approach as the “working title mindset.” In writing, a working title is useful precisely because it’s provisional. It gives you something to aim toward while you draft, but you expect to revise it once you know what you’ve actually written. The title First Impressions served Jane Austen well while she was drafting, but the book we know as Pride and Prejudice needed a different name to capture what it had become.
What would change if we treated every developmental label the same way? “Speech-delayed” would become a working title, not a final one. It would signal: “Here’s what we’re noticing right now, with this tool, in this setting. Let’s keep watching. Let’s ask more questions. Let’s be ready to revise.” The label would open inquiry rather than close it.
This mindset also changes how we design the tools themselves. A developmental screening tool built with the working title philosophy would include prompts for context: “Is this child learning more than one language?” “Who lives in the household?” “What does this skill look like in your daily routines?” It would generate not just a score but a set of questions for follow-up. It would be, like a good book title generator that helps writers find a direction without pretending to offer the final answer, a spark rather than a stamp.
I’m not suggesting we replace developmental science with vague optimism. Precise measurement matters. Early intervention matters. But precision and humility are not opposites. The most precise tools are the ones that know their own limits.
What Gets Erased When We Name Too Quickly
There’s a deeper issue here, one that goes beyond individual children and labels. When a screening tool produces a label like “delayed” for a child whose behavior is actually adaptive to their context—a bilingual child, a child in a new caregiving arrangement, a child whose family communicates through gesture and shared activity more than through direct verbal quizzing—the label doesn’t just misdescribe the child. It erases the context. It makes the child’s environment invisible and locates the “problem” inside the child.
This erasure has consequences. It can lead to interventions that target the child rather than the environment. A child labeled “delayed” in language may be referred to speech therapy, which may help, but if the real issue is that the child’s language environment isn’t being recognized by the tools, the therapy may miss the mark. Worse, the label can travel with the child into school, where it shapes teacher expectations and peer perceptions in ways that compound over time.
I’ve seen the opposite erasure, too. A child whose behavior is genuinely concerning—a child who is not making eye contact, not responding to their name, not using gestures to communicate—may be labeled “shy” or “just taking their time” by a provider who doesn’t want to alarm the family. That label, too, erases something: the opportunity for early support that could make a significant difference. The problem is not that we use labels. The problem is that we use them as if they were endpoints rather than starting points.
An Open Question
I want to end with a question rather than a conclusion, because the working title mindset requires us to stay open. What would change if we treated every developmental label as a working title—provisional, revisable, and never the whole story?
For Miriam and Leo, it might mean that the pediatrician, instead of checking a box and printing a referral, says: “Tell me how Leo talks at home. What does he say when he’s excited? What does he say when he’s frustrated? In what languages?” It might mean that the label on the form is penciled in, not inked. It might mean that Leo’s story stays open long enough for him to write the first chapter himself.
For the rest of us—parents, teachers, clinicians, writers—it might mean holding our names more lightly. A label is a tool, not a truth. A title is a direction, not a destination. The small, well-observed question—“What did you actually see?”—can lead to clearer thinking, better conversations, and more humane decisions. It can remind us that the child, like the book, is still being written.
Recent Comments