Method & Research

Nobody teaches you to read a form

A few years ago a research team gave a group of people with no legal training a task: take a simple rule and write it the way a law is written.

They all wrote it the same way — the same shape, whatever the wording. Long definitions shoved into the middle of sentences, so that the thing doing the acting and what it does end up separated by a paragraph’s worth of qualification. The technical name for this is center-embedding. When the same researchers tested which features of legal writing actually impede comprehension, center-embedding was the biggest one.

What surprised them about the writing experiment was that it showed up in first drafts, before anyone had a chance to revise. Nobody edited their way into legalese. They reached for it immediately, because they understood, without being told, that this is how a law is supposed to sound.

The finding is about native speakers writing in their own language. Hold onto it, because it changes what happens when you meet the same sentence in an official document, in a language you are still learning.

Look up every word and you still will not know what to do

The first instinct, when an official document stops you, is to blame vocabulary. It looks like a word problem. There are unfamiliar words on the page, so you look them up, and the frustrating part is what happens next: you now know every word and you still cannot say what the document wants from you. Which is strange, and worth taking seriously, because it means the words were never the whole problem.

They are part of it. The same researchers found that jargon does genuinely impede comprehension, so this is not a case of vocabulary being a red herring. But the unfamiliar words in a notice are a different species from the unfamiliar words in a novel. A novel’s hard words are usually about something — a kind of bird, a piece of furniture, a feeling. You can often work them out from what is happening around them, and if you cannot, a dictionary genuinely solves it. The hard words in administrative prose often describe the document’s own machinery: what is being granted, what is being waived, what takes effect notwithstanding what. Context cannot teach those words the way it teaches a noun, because there is no scene to attach them to.

The sentence is built to keep you waiting

Here is a specimen. It is in English, which is the point — read it in the language you know best.

Any person who, having been notified under subsection (b) and having failed within thirty days of such notification to file the statement required thereunder, subsequently requests an extension, shall be assessed a fee.

Now the same rule, said plainly:

We notify you. You have thirty days to file the statement. If you miss that and then ask for more time, you pay a fee.

Nothing was removed. The second version contains every condition the first one does. What changed is that the second lets you finish one thought before starting the next, while the first makes you hold any person in your head, unresolved, through three separate qualifications before it finally tells you what happens to them — a fee, at the very end, after you have spent all your attention on the conditions.

That is center-embedding. Long sentences are not the problem; you can write a long one that is easy to follow. The problem is interruption, and how much you are made to carry while it runs.

Go back to the first version and watch what you do with it. You are fluent in English, and you almost certainly had to read it twice. The structure defeats native readers too; that is what the study found.

Now put it in a language you are still learning

Everything above is the baseline difficulty — what a document does to someone reading it in their first language, with a full vocabulary and a lifetime of grammatical intuition to spend.

A learner meets the same structure with less margin. Grammar is not what gives out. Holding an unresolved subject across several qualifications simply costs more when each qualification also takes a moment to parse. A native reader spends the whole budget on the structure. You are spending it on the structure and the language at once.

There is a second thing happening, and it is worth being careful about, because the obvious version of it turns out to be wrong. Administrative prose removes the person from the sentence. The action becomes a noun — in the event of non-appearance rather than if you do not appear. The agent goes missing — it shall be deemed rejected, by nobody in particular. Plain-language guidance has long blamed the passive for this, and the same researchers tested that directly and found it does not affect comprehension at all. Removing it would not make the document easier to process.

What matters is what gets left unsaid. Conversation almost always has a person as its subject, and if your reading has been built on conversation and on stories, you have spent years training on sentences that tell you who is doing what. A notice often takes that away — not because the grammar is harder, but because the fact is simply absent from the page.

So you arrive at a document that does not say who must act, wraps the rule around its exceptions, and expects you to extract an obligation from it.

It was not written to be understood by you

A news article is written for a stranger. It assumes you arrived with nothing and it supplies what you need as it goes. An administrative document assumes the opposite. It refers to a deadline it never states, cites a rule by its number and moves on, and says the aforementioned period about a period nobody mentioned in your presence.

The obscurity is not decorative. The document is doing its job, which is to be a record inside a process, addressed to a file rather than to a reader.

The document is not a bad explanation. It is not an explanation at all.

Reading one is a different act

Which means you should stop trying to read these documents the way you read everything else.

You do not read a form for comprehension. You read it for three things — what you must do, by when, and what happens if you do not — and almost everything else on the page is there for somebody other than you. The clauses about other categories of person, the citations, the qualifications that turn out not to apply: you are meant to skip those, and skipping them is most of the technique.

Extraction is a skill of its own, and it has its own moves. Find the verb that attaches to you. Find the number that is a deadline. Read the sentence backwards from the consequence to the condition, because the consequence is usually stranded at the end. Decide what is addressed to your situation and let the rest go unread.

Nobody assigns that. No course has a unit on it, no textbook chapter covers it, and most people meet their first real document at a bank counter or a housing office with the clock running.

Where you would get the practice

The problem is that you do not get much practice with this kind of reading. Look at what you have actually read in the language so far: stories and conversations, almost certainly, and very little else. So the first register you meet outside a book may well be the one you have never once met on a page.

Reading widely helps. Reading across registers helps more, and that takes material deliberately built to include the documents alongside the conversations. Building it is what we do at Omorphos Publishing, and the method behind it is written up in full.

If the plateau itself is the part you recognize — conversation working, real reading still out of reach — that argument is here.

Comments