Why English? | Naming What an Image Button Does Before Describing How It Looks

A worksheet page has a small picture of a paper sheet. One student calls it “the blue document.” Another calls it “the thing on the left.” Neither phrase tells a reader whether activating it opens instructions, downloads a worksheet or sends an answer. The picture is visible; the action is still unclear.

Image-button alt text, descriptive link text and accessible control names are practical English writing tasks. A useful label explains the action or destination that matters to the reader. Describing an icon’s colour and outline may be accurate while failing to explain what the control does.

This guide uses an original fictional learning page called Practice Desk. Its controls and behaviours are supplied as a text-based workshop, not a functioning website, an accessibility audit or a record of a real school platform. World Wide Web Consortium guidance was checked on 18 October 2026. Technical implementation still needs inspection by someone responsible for the actual product.

The English goal is narrow: choose action words and objects that let a reader understand a control’s purpose. It is not a general guide to describing photographs, writing science-diagram alternatives or translating interfaces. English helps when it is shared by the people using the page; thoughtful naming matters in other languages too.

Start with the action the reader is choosing

Look at a fictional picture of a downward arrow. You could describe it as “a blue arrow pointing down.” That might help someone imagine the shape. But if the picture is a control, the reader also needs to know what selecting it will do.

The arrow might download a file, move to a later section, open a menu or change an order. Its shape alone cannot settle which action applies. A useful name starts from the verified behaviour, not from a confident guess about what arrows normally mean.

The W3C Web Accessibility Initiative’s functional-images guidance explains that text alternatives for images used to initiate actions should communicate their function rather than merely their appearance. That principle gives our English workshop its central question: “What is the reader being invited to do here?”

Once you know the action, identify its object. “Download” tells us a kind of operation. “Download blank worksheet” tells us which resource. “Download blank worksheet (PDF)” adds a relevant format when that format is known. Each extra word should improve the reader’s decision, not simply make the label longer.

This does not mean every control needs a full sentence. Short language can be exact. “Save draft” is only two words, but the verb and noun distinguish it from reviewing, sending or opening a document. The test is useful meaning, not an arbitrary length target.

Before polishing a name, ask the product owner or teacher to explain the actual behaviour. If that information is missing, mark the question for clarification. You cannot repair an unknown action merely by finding a more attractive adjective for the icon.

Back to contents

Meet the fictional Practice Desk page

Our invented page lets a learner prepare a short response about a book. The exercise supplies the following behaviour descriptions. They are the evidence for choosing names within this workshop; they are not statements about a real learning-management system.

Control	Supplied behaviour	Proposed action name
A: paper-sheet image	Opens instructions in the current page	Open instructions
B: downward-arrow image	Downloads a blank worksheet as a PDF	Download blank worksheet (PDF)
C: disk-shaped image	Saves the current response as a draft; does not send it	Save draft
D: eye-shaped image	Opens a review screen; does not send the response	Review before sending
E: final control on that review screen	Sends the response when activated	Send answer

The table separates appearance, behaviour and name. Appearance helps us discuss which fictional control we mean. Behaviour tells us what happens. The action name gives the reader a concise way to choose among those behaviours.

Notice that the eye image is not named “View answer key.” The supplied behaviour says review screen, not answer key. A familiar symbol does not authorise us to invent a destination. Likewise, the disk-shaped image does not establish submission; the supplied behaviour explicitly says draft only.

For a classroom version, write the behaviours on separate cards before learners see any icons. Ask them to propose names from those cards. Then show the images and ask whether the names still preserve the action. This keeps visual familiarity from doing all the reasoning.

Our workshop stops at language and interpretation. It does not test keyboard access, focus order, browser announcements, error handling or whether saved work really persists. Those are important product questions, but an invented text table cannot provide evidence that they have been implemented correctly.

Back to contents

The verb carries the kind of action

Compare open, download, save, review and send. They are all ordinary English verbs, but they describe different relationships between the user and the material. Replacing them with the vague word “go” would remove much of the information a reader needs.

In our supplied case, open makes instructions available on the current page. Download obtains the worksheet file. Save stores a draft according to the fictional behaviour. Review moves to a checking screen. Send performs the final transmission action. None should be used as a casual synonym for all the others.

Try the label “Finish” for control D. It sounds encouraging, but the control does not finish sending. It opens a review screen. A learner who reads Finish as submission may stop before using E. A pleasant word can therefore create an inaccurate account of the workflow.

“Continue to review” is another possible name for D if it suits the surrounding page. “Review before sending” makes the sequence particularly explicit in this exercise. The best wording depends on context; the necessary condition is that it does not claim that sending already happens at D.

Verb tense also matters in feedback. “Send answer” names an available action. “Answer sent” reports a completed state. Writing the second phrase on an action control could suggest completion before the action has occurred. In a real interface, the wording must be checked against the actual state and behaviour.

A student can practise by sorting phrases into action names and status messages. Ask why “Draft saved” belongs in a different category from “Save draft.” The change is small, but it tells the reader whether something can be done or is reported as having happened.

Back to contents

The object tells the reader what the action affects

“Open file” can be grammatical and still leave the reader with an important question: which file? Our Practice Desk page has instructions, a blank worksheet and a response. Naming the object helps people distinguish among resources before they activate a control.

For B, the word blank is useful because it separates a reusable worksheet from a completed response or answer key. We know the file is blank because the fictional behaviour supplies that fact. Without that evidence, adding blank would be an unverified promise.

For C, draft is useful because it identifies a working response rather than a final submission. It does not mean that saving is unimportant. It means the reader should understand which stage the operation concerns and which later action remains available.

Now imagine two controls named “Download.” One obtains instructions, while the other obtains the worksheet. Their verb is correct, but the pair is difficult to distinguish. “Download instructions (PDF)” and “Download blank worksheet (PDF)” would preserve the difference if those are the verified resources and formats.

Avoid putting personal data into a label simply to make it more specific. Our exercise needs no student names or account identifiers. In real work, use the product’s appropriate naming and privacy conventions, and do not expose private information where a task-based description would do.

The grammatical pattern is simple: action plus meaningful object, with a short qualifier when it changes the decision. The reasoning behind the pattern is more important than memorising it. Ask which uncertainty each word removes. If a word adds no useful distinction, it may not belong in a compact control name.

Back to contents

An image-only control and an image beside text are different cases

Suppose A contains only a paper-sheet image. A reader who does not receive that image’s visual information still needs the control’s purpose. The image’s text alternative can carry that function when the image is the only content of a link.

Now suppose the same link contains both the image and visible text saying “Open instructions.” If the image adds no further information, repeating those words as its alternative may be unnecessary. W3C’s H30 technique for descriptive link text distinguishes these situations and describes an empty image alternative where sufficient link text already supplies the purpose.

Do not turn that conditional example into “all button images should have empty alt text.” The condition matters: the control’s purpose is already provided by the text, and the image contributes no extra information. Removing the only useful name from an image-only control would be a different situation.

Equally, do not assume that any nearby paragraph becomes part of a control’s accessible name. Actual relationships depend on implementation. Our text workshop can judge whether a proposed phrase communicates the intended action; it cannot confirm what a browser or assistive technology will expose on an uninspected page.

The writer and implementer should cooperate. The writer can provide the intended action, necessary distinctions and proposed wording. The implementer can establish how the real control receives its name and whether the result works as intended across the supported experience.

Students do not need to master every accessibility mechanism to understand the language distinction. They can learn to ask whether the purpose has already been supplied, whether an image adds something new and whether a proposed name leaves the reader with the same useful choice.

Back to contents

Use visible wording and accessible wording as one coherent explanation

Imagine that the visible text says “Save draft,” while a separate supplied name says “Send answer.” The two phrases describe different actions. Different readers could reasonably expect different outcomes from the same control, even before we inspect any technical implementation.

The first repair is to establish what the control actually does. In our case C saves a draft only. The intended language should preserve that behaviour for everyone. A richer explanation may be useful, but it should not quietly contradict the action named by the visible words.

This matters in collaboration as well. A teacher may tell students to choose Save draft. If one learner encounters a name about sending, the instruction and control no longer align. The learner should not have to guess that contradictory words are meant to describe the same operation.

Consistency is not the same as mechanically repeating identical sentences everywhere. A short control can say “Save draft,” while nearby help explains that the response is not sent yet. Those two pieces reinforce one another. Calling the control “Submit final response” would change the meaning rather than expand it.

When handing wording to a developer, write the behaviour beside the proposed name. For example: “C stores a working draft without sending; proposed name: Save draft.” This gives the implementer something testable and makes a mismatch easier to spot during review.

An actual accessibility evaluation must consider more than wording, and this article does not claim conformance with any standard. Still, semantic consistency is a worthwhile contribution a learner can make: the words should tell a coherent story about the action the product really offers.

Back to contents

A review screen is not a sent answer

Follow our fictional sequence carefully. A learner writes a response, activates C and then activates D. According to the supplied behaviours, the draft has been saved and a review screen has opened. E, the final sending action, has not yet been activated.

The sentence “I reviewed it, so I sent it” skips a step. Review can support a decision to send, but the two actions are not identical. Our control name “Review before sending” makes the distinction visible without adding a warning to every sentence.

Now consider the message “Ready to send.” It describes a preparatory state, not completion. “Sending” describes a process under way. “Sent” reports completion. In a real system, each statement needs appropriate evidence from the actual workflow rather than inference from a familiar-looking icon.

This workshop does not prescribe how a real platform must confirm delivery. It shows why readers should not convert an earlier action into a later one. If schoolwork matters, follow the teacher’s current submission instructions and check the actual confirmation available on the authorised platform.

As a practice task, ask a partner to narrate the supplied actions without using the word done. They might say, “I saved a draft, opened the review screen and have not used Send answer.” That account is more useful than a vague completion claim because it names the current boundary.

The point is not to make ordinary tasks feel fragile. It is to make the next move easy to see. Once the learner can distinguish saving, reviewing and sending, the workflow becomes less mysterious and a question to the teacher can be much more precise.

Back to contents

When one symbol appears in two different jobs

Invent a second page with two paper-sheet images. One opens the activity instructions; the other opens the learner’s own saved response. Their shapes match, but their destinations differ. Naming both controls “Document” would erase that distinction.

Appropriate names might be “Open activity instructions” and “Open saved response,” provided those behaviours are actually supplied. The noun phrase does the separating work. No amount of detail about the shared page outline can tell a reader which resource lies behind each control.

Now change the example. A magnifying-glass image appears beside a search field, but another magnifying-glass image enlarges a diagram. “Search activities” and “Enlarge diagram” name different actions. We should not choose one label merely because it is the most familiar interpretation of the symbol.

These are original hypothetical cases, not claims about a particular product’s icons. They demonstrate a general writing method: first ask what this control does here, then choose words. A symbol’s common association may help a reader, but it is not a substitute for the verified local behaviour.

This also explains why a picture-description exercise and a control-naming exercise need different success tests. “A sheet with a folded corner” can be a useful visual description. For choosing a document action, it may leave the central question unanswered. The reader’s task determines what information the words need to carry.

For the spoken-description task, use the existing guide to describing a picture to someone who cannot see it. Its purpose is different from naming the action of an interactive image.

Back to contents

Did you know? A beautiful description can be the wrong kind of answer

Consider the phrase “A polished silver disk with a small square cut-out.” It might describe the fictional image precisely. It still does not tell the learner whether the control saves a draft, opens a file or performs some other action.

That is a useful discovery about English: accuracy depends on the question. A visual-description question asks what something looks like. A control-purpose question asks what someone can do with it. The same fluent sentence can succeed at the first job and fail at the second.

This is why a writer should not begin by searching for grander adjectives. Begin by identifying the missing relationship. In our main case, the missing relationship is between the user’s choice and its result. A plain verb such as save can supply more useful meaning than several descriptive phrases.

Concise wording is not automatically good either. “Go” is short, but vague. “Process” may sound professional, but does not distinguish opening, saving, checking or sending. The appropriate economy is to remove words that do not help while retaining words that establish the action and object.

Try a small comparison exercise. Give someone the five appearance descriptions from our table, then ask which control sends the answer. Without the supplied behaviours, they should not claim certainty. Now give them the action names. The decision becomes much clearer because the language carries the relevant information.

The reward is practical cooperation. Readers can choose, teachers can give instructions and teammates can report progress using the same task words. This does not eliminate the need to test a real interface, but it gives that test a clear intended meaning to compare against.

Back to contents

A naming workshop with reasons, not just preferred words

Here are four proposed names for B: “Arrow,” “Get it,” “Download answers” and “Download blank worksheet (PDF).” The supplied behaviour is downloading the blank worksheet as a PDF. Only the fourth proposal preserves both the action and the verified resource without adding a different promise.

Arrow describes appearance. Get it leaves the object uncertain. Download answers invents answer content. The fourth is not better because it is longest; it is better because its extra words resolve real differences relevant to the reader’s choice.

For D, compare “Preview,” “Review before sending” and “Answer sent.” Preview could work in a context where the object and next step are already clear, but in our standalone exercise the second phrase is more informative. The third is wrong because D does not send anything.

For C, compare “Keep,” “Save draft” and “Store response forever.” Keep may be too vague. Save draft matches the supplied behaviour. Forever makes an unsupported retention promise. Even when a writer wants to reassure a learner, they should not add a guarantee absent from the actual product information.

Ask students to justify each decision with the pattern “This wording implies ___, but the supplied behaviour is ___.” The explanation exposes a semantic mismatch and gives the writer something concrete to change. It is more teachable than declaring that a label simply sounds wrong.

Then ask whether surrounding context could make a shorter name sufficient. The goal is flexible judgement, not one rigid label for every product. Preserve the action and necessary distinctions, check the real context and keep any implementation claims within what has actually been verified.

Back to contents

Parent advice: make the activity about choices, not disability assumptions

You can run this exercise using paper cards without asking a child to simulate another person’s disability. Give the learner a control’s intended behaviour and ask for a useful name. Discuss what a reader would need to choose the correct action without relying on the picture.

Avoid assuming that everyone uses a website in the same way or wants the same amount of spoken description. When working with real users, ask about their needs and follow appropriate accessibility expertise. Our workshop teaches language choices; it does not replace the knowledge or preferences of people using assistive technology.

If a child supplies “blue arrow,” ask a task question: “What would you expect to happen after choosing it?” Let them notice that the phrase does not say. Then provide the known behaviour and invite a revision. This makes the improvement understandable rather than a correction to memorise.

If the child writes a long paragraph, ask which words help distinguish the action from other available actions. They may be able to keep the essential verb and object while moving background explanation into nearby help. Concision becomes a reasoning task rather than an instruction to use fewer words at any cost.

Praise the specific improvement: “You distinguished saving a draft from sending the answer.” That is more informative than general praise for sounding clever. It tells the learner which meaning was preserved and gives them a habit to use in another task.

Keep all examples fictional unless the teacher has authorised work on a real platform. Do not change school controls, account settings or website content for practice. A paper exercise is enough to reveal the difference between an image’s appearance and the action its words need to explain.

Back to contents

Student advice: hand over meaning a teammate can implement

If your project group is designing a learning page, prepare a small wording sheet rather than sending a list of isolated labels. For each control, record the intended behaviour, proposed name, important distinction and question still requiring confirmation.

For C, that sheet could say: “Saves current response as draft; name Save draft; must not imply sending; confirm how the actual product reports a successful save.” The last item is a question, not an invented feature. It helps your teammate know what still needs testing.

For B, record that PDF is a supplied format in the design, not a guess from the arrow. If the product later changes to a different format, the name and supporting explanation need review. Words are part of the experience and can become stale when behaviour changes.

Ask someone outside the design discussion to choose a control for a simple goal. For example: “You want to check your response before sending it. Which name would you choose?” This can reveal whether your wording distinguishes the available actions, though a few informal trials do not amount to a full usability or accessibility study.

When reporting feedback, describe what happened rather than making a universal claim. “One classmate expected Review to send the response” is a useful observation if that actually occurred. “Everyone finds Review confusing” would need much stronger evidence. In our invented workshop, do not report trials as if real users completed them.

Finally, keep the action description available during later changes. A new icon should not accidentally alter the promised behaviour, and a changed behaviour should not hide behind an unchanged name. The English work is to keep the reader’s expectation aligned with the product’s intended action.

Back to contents

Transfer to learning, design and future work

This small writing task connects to a larger educational habit: explain what a person can do, what information the action affects and what happens next. That habit can help with classroom instructions, project handovers and explanations of unfamiliar digital tools.

It also shows why language learning is not limited to long essays. A two-word name may carry an important decision. The challenge is not necessarily finding advanced vocabulary; it is selecting ordinary words whose distinctions are useful to the reader.

Students interested in design, teaching, computing or communication can notice how those interests meet here. A product needs behaviour, a reader needs a clear choice and a writer needs evidence about the intended action. No single interest replaces the others, and this exercise is not a test for entry into a particular course or occupation.

For further-study decisions, use current official programme and admissions information when you reach that stage. Do not treat a successful naming activity as proof of eligibility or a promise of employment. It is one small opportunity to practise explaining a relationship that many collaborative tasks depend on.

The broader media-accessibility guide covers other routes through captions, transcripts, images and inclusive access. The existing article on alt text for science diagrams owns the different task of preserving information and relationships in educational visuals.

Our contribution is more focused: when a picture functions as a control, name the action rather than stopping at appearance. It is an encouraging example of English helping a real choice become easier to understand, without requiring a dramatic sentence or an impressive collection of technical terms.

Back to contents

Independent practice: three new control briefs

Brief one describes a fictional headphone image that opens a transcript, not audio. The designer chose an unsuitable-looking icon, but the supplied behaviour is still opening the transcript. “Play recording” would misdescribe that behaviour. A truthful proposed name is “Open transcript,” alongside feedback that the visual cue also needs review.

This example prevents a common shortcut: assuming that naming can be completed from appearance alone. Sometimes a clearer label exposes a problem in the design itself. The writer should not fix the contradiction by inventing an action the control does not perform.

Brief two describes a folder image that opens a list of saved drafts. “Open saved drafts” carries the destination. “Create folder” would imply an action absent from the brief. “Folder picture” describes the illustration but leaves the reader’s choice unresolved.

Brief three describes a tick image that checks spelling in the current draft but does not judge factual accuracy or send work. “Check spelling” is appropriate to that supplied scope. “Check everything” overpromises, while “Submit” names an entirely different action. A nearby explanation can clarify limitations without forcing every detail into the compact name.

Now ask a partner to produce one deliberately misleading name and explain the false expectation it creates. The value lies in the reason, not in tricking someone. If the group can identify the unsupported action or object, it has something specific to repair.

Finish by returning to Practice Desk. Without looking at the icon descriptions, explain the difference between Save draft, Review before sending and Send answer. If you can name what each does and what it does not establish, you have practised the central language skill rather than merely memorised a picture.

Check a changed brief without carrying the old promise forward

Suppose the fictional designer changes B. It no longer downloads a PDF; it opens an editable worksheet on the current page. The arrow picture remains unchanged. The old name “Download blank worksheet (PDF)” is now misleading even though nobody has altered a word in the label.

Review the behaviour first. A possible new name is “Open editable worksheet,” provided the new brief actually supplies that behaviour. Do not keep PDF because it was true in an earlier version, and do not add automatic saving unless that has also been established. The wording should describe the current action, not the designer’s remembered intention.

This is a separate revision branch, not a change to our original Practice Desk table. In that original case, B still downloads the blank PDF. Keeping the two cases distinct prevents a new example from silently rewriting the evidence used in earlier exercises.

Ask students to produce a small change note: “B now opens an editable worksheet instead of downloading the blank PDF; update the action name and recheck related help.” That note gives the next teammate a specific relationship to inspect. It also shows why English remains useful after a page is first written: words must continue to match behaviour as a project develops.

Back to contents

Frequently asked questions

Should alt text always describe the picture’s appearance?

No. The useful information depends on the image’s role. W3C distinguishes functional images from images that communicate information. In this article’s narrow case, the picture is part of an action control, so its purpose matters. A separate photographic or science-diagram description may need a different kind of text.

Can an image beside sufficient link text have an empty alternative?

W3C describes that situation when the link text already supplies the purpose and the image adds no further information. Do not apply the example mechanically to image-only controls. The actual implementation needs review so that the control still has a useful name and any extra information is preserved.

Is “click here” a good action name?

It does not identify the action or destination on its own. In our workshop, “Open instructions” or “Download blank worksheet” gives the reader more relevant information. Write for the choice being offered, rather than assuming that every reader uses the same pointing device or can infer a destination from visual position.

Does Save draft mean my teacher has received the answer?

Not in the supplied fictional workflow. C saves a draft; D opens review; E sends. Real platforms have their own behaviours and confirmation messages. Follow your teacher’s current instructions and inspect the actual state rather than importing this invented sequence into a different product.

Can a good label make an entire website accessible?

No. Meaningful names are one part of a wider experience that also needs appropriate implementation and testing. This article does not audit a website, test assistive technology or establish conformance. It teaches a specific English-writing contribution: preserving the intended action in the words a reader receives.

What if I do not know what a control does?

Ask the responsible person or consult verified product information before naming it. Treat an unverified interpretation as a question, not a finished label. A convincing sentence cannot supply missing behaviour evidence. In a student project, a clear unresolved question is a useful handover item.

Back to contents

Useful next reading

Continue through Why English Matters or the English Learning Hub for broader learning routes. SETC’s How English Works explains how grammar, meaning and context support useful language choices.

Use the official W3C links above for the documented functional-image and link-text distinctions. For collaborative production, the guide to school projects and shared presentation work provides the wider context. Keep the action small, truthful and clear: tell the reader what choosing the control is intended to do.

Back to contents