Begin with the information and the person’s task
Choosing a device is easier when the requirement says what a person needs to do. “Use a computer” is too broad. “Enter a short equipment note without using hands, then check every number” identifies the input, the output and an important accuracy check. A good answer explains why the complete method fits, identifies a limitation and supplies a usable fallback.
The following adult/general scenarios are fictional. Preferences and capabilities are supplied facts for the exercise, not diagnoses or assumptions about a group. No equipment purchase or real personal information is needed. Work through each choice on paper.
Separate the hardware from the method
An input device supplies information or control signals to a system. An output device presents information from it. A microphone captures sound; speech-recognition software is needed to turn spoken words into proposed text. Loudspeakers reproduce sound; text-to-speech software can generate spoken output from text. Naming only the hardware leaves the conversion unexplained.
Classification follows the function being used. A touch display provides visual output while its touch sensor receives input. Touching a displayed button is input; seeing the confirmation is output. A screen reader is software rather than a speaker. It needs usable content and controls, and it cannot reliably supply every important detail that an author omitted from an image.
| Task requirement | Input method | Output and software needs | Constraint to check |
|---|---|---|---|
| Write and edit several sentences | Physical keyboard | Text editor and visible editable text | Can the person reach and use the required controls? |
| Create text without using hands | Microphone with speech recognition | Recognised text plus accessible review and correction | Background sound and supported commands |
| Choose from a short menu | Touch controls or keyboard selection | Clear labels and visible confirmation | Target size, keyboard access and user preference |
| Understand a visual status summary | No new input necessarily required | Structured text, screen reader and audio device if speech is chosen | Meaning must exist in interpretable content |
Worked case 1: typed text in a quiet workspace
An adult organiser must enter “Return the three display stands before noon” and later change “three” to “four”. The supplied facts say they comfortably use a keyboard, prefer quiet working and can review displayed text. The final requirement is editable text, not an audio recording.
Choose a keyboard, a text-editing application and a display. The keyboard permits direct character entry and the application supports moving through and replacing text. A pointing device may help select text, but it is not automatically essential if the relevant editing controls support keyboard use.
The observable check is the complete revised sentence: “Return the four display stands before noon”. Check the changed word and the unchanged deadline. Dictation is a possible alternative, but speaking adds an unnecessary interaction for this stated preference. If a control cannot be reached by keyboard, the choice is incomplete: use an agreed compatible editing route or a pointing method the person can use. Do not simply declare the person unsuitable.
Worked case 2: hands-free text with a noisy-room limitation
A demonstration assistant needs a hands-free way to enter “Box seven needs two clips”. The assistant is comfortable speaking and reviewing spoken text. The room currently contains loud unrelated conversation. A microphone alone would capture sound rather than produce the required text.
Choose a microphone with compatible speech-recognition software and a review method, such as text-to-speech playback. Also check that correction and confirmation controls can be operated without hands. Dictating the sentence is only part of the job; a system that requires an unusable mouse action to correct it does not meet the whole requirement.
Suppose the fictional proposed transcript is “Box eleven needs two clips”. It is grammatically plausible but wrong. Compare it with the supplied intended sentence, correct “eleven” to “seven”, then review the whole sentence again. Better punctuation would not fix the wrong box number.
Background noise may reduce recognition reliability. A quieter location or a pause until the conversation ends is the agreed fallback here. A different microphone might help in some conditions, but it is not a guarantee. Typing is not a valid automatic fallback when this case explicitly requires hands-free interaction. If no workable correction route exists, pause the entry rather than accepting unverified text.
Worked case 3: visual information with equivalent output
A fictional display shows two store zones: North has six stands ready and South has two. One adult reader prefers speech; another prefers visible text. A picture with two coloured blocks and no labels would hide both the exact quantities and the meaning of the colours.
Provide a structured text equivalent: “Stands ready: North, 6; South, 2; total, 8.” Retain visible labels beside the diagram and offer content that a screen reader can interpret. Speech through headphones is one possible output route when the listener wants it; a visible text table is another. No person is required to use both.
Check equivalence by asking the same questions of each representation: which zone has more, how many are ready in South, and what is the total? The answers are North, 2 and 8. A description saying only “stand chart” fails this check. If audio is unavailable, the text remains useful for readers who can use it; a different reader may need another agreed format. Accessibility is a match between content, interaction and the individual task.
A choice you can defend
Write an answer in five parts: required information, supplied preference, complete hardware/software route, limitation and check. Compare alternatives against the same requirement. A faster method that makes correction impossible is not necessarily better. A larger collection of devices is not evidence of a better fit either.
Keep the fallback realistic. It must preserve the important requirement, such as quiet work or hands-free correction. Where the facts do not establish a suitable alternative, say what needs checking instead of inventing a capability. A paper comparison can explain the proposed fit, but actual use still needs verification in the intended environment.
Three solved practice tasks
Task 1: classify a touch display
A person taps “Next” and the display shows the next page. Identify both functions. Solution: sensing the tap is input; presenting the new page is output. Calling the whole device only an input device misses its display function.
Task 2: repair an incomplete recommendation
A proposal says, “Buy a microphone so spoken notes become typed text.” What is missing? Solution: compatible recognition software, a way to review its proposed text and a usable correction route. The hardware captures sound but does not by itself establish accurate transcription.
Task 3: preserve the information
A graphic says Zone East has four trays and Zone West has five. Its alternative says “two zones”. Improve it. Solution: “Trays available: East, 4; West, 5; total, 9.” This preserves the labels, quantities and comparison without depending on colour.
Choose persistent storage against the whole task
Add a storage step to the input/output tasks above: carry the two LDP01 demonstration PDFs to an offline workstation, keep them after normal power-off and open the unchanged learner and answer versions there. The actual prepared files occupy 102,626 + 49,254 = 151,880 bytes. Keep a further 20,000 bytes free in this paper model, so the minimum free capacity is 171,880 bytes. No new copy, purchase or device action is required by the exercise.
Reuse the hardware distinction: RAM holds active working data, while persistent storage holds saved files without continuous operating power. A visible preview is output; it is not proof of a completed save. These supplied device cards are fictional constraints, not specifications or prices for actual products.
| Card | Free bytes | Persistence and access facts | Decision |
|---|---|---|---|
| A: removable flash storage | 150,000 | Persists; compatible port and readable filesystem | Fails capacity by 21,880 including reserve |
| B: removable flash storage | 200,000 | Persists; compatible port and readable filesystem | Meets every supplied constraint |
| C: external hard drive | 1,000,000 | Persists; its only connector is unsupported by the workstation; no adapter supplied | Fails compatibility despite ample capacity |
| D: RAM working area | 2,000,000 | Loses working contents at power-off; no autosave or recovery here | Fails persistence |
| E: remote storage service | 1,000,000 | Files available only over a network in this supplied card; destination is offline | Fails access; no offline copy supplied |
Worked selection: B has 200,000 − 151,880 = 48,120 bytes left after storing the pair. Reserving 20,000 leaves 28,120 bytes of spare margin. A cannot hold even the pair: it is 1,880 bytes short before the reserve is considered. C’s larger number does not repair its connection mismatch. B’s supplied fit establishes a candidate for this task, not immunity from failure or proof that every flash device works.
| State / test | Stored result or expected observation | Decision |
|---|---|---|
| S0: before copy | No LDP01 pair on B; 200,000 B free | Two source files and their versions identified |
| S1: copy learner completes | Learner 102,626 B present; 97,374 B free | Guide still required |
| S2: copy guide completes | Both files total 151,880 B; 48,120 B free | 20,000 B reserve met |
| S3: compare the supplied source and stored copies | Same version IDs, bytes and complete contents in this successful paper case | Do not infer correctness from filenames alone |
| S4: disconnect by the permitted procedure, then normal power-off | Persistent files remain in the model; RAM state is not used as evidence | No file deletion assumed |
| S5: reopen on the compatible offline destination | Both PDFs open; learner T1–T4 and corresponding answer sections present | Transfer task passes under the stated conditions |
| Edge: guide copy stops before completion | S2’s complete guide is absent; expected total and match fail | Do not report a complete pair; preserve originals and retry only under authorised real conditions |
| Edge: exactly 171,880 B initially free | After the full pair, 20,000 B remains | Passes the capacity boundary exactly |
| Edge: destination application cannot open a PDF | Bytes may still match, but usable output is missing | Task fails until a compatible authorised reader is available |
For the actual input/output accessibility job, retain the user’s stated review and correction route. A storage device does not provide text-to-speech, an accessible PDF structure or hands-free controls. A successful paper capacity check must be combined with the earlier complete input/output method and an actual access check before claiming suitability.
Independent changed-file task
Task: in a separate hypothetical revision, the guide grows by 30,000 bytes. Does B still meet the unchanged reserve? Answer: the pair would use 181,880 bytes and leave 18,120; that is 1,880 short of the 20,000 reserve. It fits physically but fails the complete requirement. Those hypothetical sizes do not replace the measured v1.0 artifact inventory.
Frequently asked questions
Is a microphone a speech-recognition system?
No. It captures sound. Recognition software interprets that input and produces proposed text or commands that still need checking.
Can one device provide both input and output?
Yes. A touch display presents information and detects touches. Classify each function rather than forcing the whole device into one category.
Does a screen reader understand every picture?
No. Important visual meaning needs an appropriate description or equivalent structured content. Merely including an image is not enough.
Is speech always faster than typing?
No. Noise, vocabulary, correction effort, compatibility and user preference affect the result. This lesson makes no universal speed claim.
Must everyone use the same output format?
No. Different people and situations call for different formats. Alternatives should preserve the information needed for the task.
Why check recognised numbers particularly carefully?
A plausible word can still identify the wrong item or quantity. Compare the entire output with the intended information, including numbers and labels.
Does this exercise recommend a particular product?
No. It explains a method for comparing task requirements with interaction methods. It makes no price, performance or individual suitability promise.
Further reading
Technical background: W3C Web Accessibility Initiative — input methods; W3C Web Accessibility Initiative — perceiving content. Examples, questions and diagrams in this lesson are original.
WhatsApp Us Now