Unicode and Accessible Social Text

Use the post readability checker for counts, located findings, and a comparison with your own revision. Then inspect the destination separately. Our worked example is a fictional museum notice, where a clear entrance instruction matters more than decorative emphasis and where an intentional emoji sequence must survive the edit intact.

Separate a writing system from a visual effect

Start by asking what each unusual element does. An accented letter can be part of a person or place name. A symbol can identify a measurement. Another script can be the correct writing system for the audience. None should be removed merely because it looks unfamiliar to an English-speaking editor. The aim is a dependable message in the intended language, not a version limited to the characters on one keyboard.

Decoration is a different choice. A social post may replace ordinary letters with mathematical bold or script characters to imitate formatting. That replacement changes the underlying characters; it is not the same as applying a font to ordinary text. The distinction matters when someone copies a name, searches for a phrase, or uses assistive technology. Do not assume that visual similarity means two pieces of text behave identically everywhere.

For example, ordinary capital A is U+0041, while the mathematical bold character 𝐀 is U+1D400. The U+ notation identifies a Unicode code point. In a museum notice, those details can help an editor recognize why a styled heading is not just ordinary text with bold presentation. They do not prove that a particular reader cannot use it. Keep the claim narrow and test the experience that actually matters.

A practical editorial default is ordinary letters for essential instructions and exact original spelling for names. Use a separate design treatment when appropriate rather than substituting a decorative alphabet across the whole message. Keep a copy of the source before changing anything. If a specialist name, mathematical expression, or quoted work uses unusual characters intentionally, ask its owner before replacing them with a visually similar but potentially different form.

Know which kind of character a count describes

Character is a convenient everyday word, but different counts answer different questions. A code point identifies an encoded character. A UTF-16 unit is part of the representation used by JavaScript strings. A grapheme cluster is a grouping used to approximate a user-perceived character. A single visible item can therefore contribute different amounts to these counts. Read the label before treating two numbers as if they measured the same thing.

The capital A example has one code point and one UTF-16 unit. Mathematical bold capital A still has one code point, but it uses two UTF-16 units. An accented letter can sometimes be represented as one precomposed code point or as a letter plus a combining mark. Both may look alike. A different underlying count is not proof of an extra visible letter, corruption, or a readability problem that needs fixing.

The checker uses Intl.Segmenter for word, sentence, and grapheme segmentation. Its input capacity is 20,000 UTF-16 units per draft, not 20,000 visible symbols. Its general word count excludes recognized HTTP URLs and includes wordlike segments with letters or numbers. Paragraphs use blank-line boundaries. These are mechanical rules for inspecting a supplied string. They do not measure how a museum visitor will interpret a symbol or pronounce a place name.

When counts change after an edit, identify the reason before deciding whether the change helps. Replacing decorated letters can alter the underlying length without changing the number of visible letters. Replacing a symbol with explanatory words can add length and clarify an instruction. Neither direction is automatically better. Keep the relevant labels in review notes, and use the character counter for a separate destination-focused length check when needed.

Review decorated alphabets as a deliberate choice

A heading can look ordinary at a glance even when its letters come from several character sets. Review any source copied from a decorative text generator, artwork file, or old campaign document. Do not rely on appearance alone to identify a person, product, or venue. For essential wording, compare with the approved plain-text name. A character that resembles a Latin letter may belong to another script and should not be casually substituted.

Assistive technology behavior varies with software, language, settings, and context. A decorated heading may be announced differently from ordinary wording, or the reading may become awkward. Avoid the blanket claim that every unusual character is silent or that every screen reader handles it identically. Test the actual destination and text when this matters. The checker cannot reproduce every combination, and its text flags are not a complete inventory of decorated alphabets.

Prefer direct wording when decoration carries the main instruction. For a museum entrance change, Use the east entrance should remain understandable without a special alphabet, repeated symbols, or a visual pattern. If an optional decorative heading stays, make sure it does not replace the practical instruction. That decision is about the visitor task. It does not require erasing expressive language or treating all typography as an obstacle.

Do not apply a universal cleanup that converts every non-ASCII character. It can damage names, scripts, and distinctions that matter. Unicode normalization is also not a general plain-language rewrite: some forms make equivalent representations consistent, while compatibility forms can remove distinctions. The checker does not automatically normalize or restyle your draft. If another workflow changes characters, compare its output with the original and obtain approval for any change to meaningful content.

Investigate invisible spaces and direction controls

Some characters affect spacing or text behavior without displaying a visible mark. A normal space, a no-break space, and a zero-width space do different jobs. U+00A0 can keep adjacent items together at a line break. U+200B can create a break opportunity without a visible gap. Finding one is a reason to inspect its location and purpose, not evidence that someone has inserted harmful text or tried to deceive a reader.

Copying from a formatted document can bring characters the author did not expect. In a simple English notice, an unexplained zero-width space inside an entrance name may be accidental. Confirm the intended wording, then edit that specific occurrence in the revision. Do not run a remove-all-invisible-characters action across the source. Keep the original for comparison, and check adjacent words after the change so an attempted repair does not join two separate words.

Bidirectional controls help manage text containing left-to-right and right-to-left writing. They can affect how nearby punctuation or numbers are displayed. Some are legitimate parts of mixed-direction plain text. If you cannot explain a finding, ask someone who understands the language and inspect the actual destination. Blind removal can make the reading order worse. A Latin-letter name beside another script is a context to review carefully, not a reason to delete that script.

The tool identifies explicit zero-width, bidirectional, and control characters with a location and excerpt, while avoiding a generic warning for an emoji joiner inside its sequence. It does not decide whether every listed control is necessary or malicious. Use the finding as an address for investigation. After a careful edit, inspect the complete sentence, nearby numbers, and any name or identifier that could be affected by the change.

Preserve complete emoji sequences

A single emoji image can come from several code points. A zero-width joiner, U+200D, can connect components into a joined sequence. For example, 👩‍🔧 can be formed from WOMAN, ZERO WIDTH JOINER, and WRENCH. Removing the joiner can produce separate symbols instead of the intended combined depiction. That is why a general instruction to delete every hidden character would be unsafe for text editing.

Other emoji sequences can include variation selectors or skin-tone modifiers. Do not assume that an invisible element or a component you cannot select visually is spare material. If an emoji is appropriate for the message, keep the complete sequence. If it is unnecessary, remove the whole intended emoji deliberately and compare the surrounding text. Do not edit its internal pieces simply to make an underlying count smaller or to clear an unrelated warning.

Decide what information the emoji contributes. A museum notice should state its opening hours and entrance in words rather than asking a symbol to carry those facts alone. An optional emoji can remain a tone choice without becoming an instruction. Listen for how it interrupts the reading flow in the actual post. Repeated decorative runs may add effort for some readers, but this is a manual judgment, not a numerical limit supplied by the checker.

Rendering can vary between devices, fonts, and software versions. A sequence may appear as one image in one setting and separate components in another. Check the intended destination instead of trusting only the review field. A count of one grapheme cluster also does not predict its spoken name or how long it takes to hear. The tool does not listen to screen reader output, assess emoji meaning, or certify the sequence as accessible.

Check hashtag casing and link context manually

A hashtag combines a visible label with a searchable form. When its intended phrase has several words, meaningful capitals can make the word boundaries easier to inspect. #MuseumLateNight presents the intended phrase more clearly than a single run of lowercase letters for many readers. That is a casing choice, not a guarantee of pronunciation or discovery. Confirm the intended campaign tag before changing its spelling or replacing it with another phrase.

The checker marks single-case hashtags for manual review. It does not infer whether a tag contains one word, several words, a name, or an acronym. A mixed-case tag can still have unhelpful boundaries, and a lowercase one-word tag can be intentional. Read the phrase yourself. Keep tags separate from essential instructions so a visitor can learn the entrance and opening hours without needing to decode a campaign label first.

Links also require human context. Tell the reader what the destination supplies, such as entrance details or a booking page, instead of relying on a bare address to explain itself. A URL in the checker is a manual review item, not a fetched resource. The tool cannot establish whether the page is available, whether a redirect leads elsewhere, or whether the destination information agrees with the wording you supplied.

If an address looks unusual, compare it with the approved destination through your normal review process. Do not replace characters inside a URL, account name, or identifier just because they resemble ordinary letters. That can change what the string refers to. Likewise, changing spacing or punctuation around a pasted address can affect what a destination recognizes. Review the final clickable result in the native composer rather than treating a clean text review as proof that every reference works.

Work through a museum evening notice

Imagine a fictional museum preparing an evening-opening notice. The approved facts are Friday opening from 18:00 to 21:00, entry through the east entrance, and an intended tag of Museum Late Night. The visible source reads: LATE OPENING!!! Friday 18:00 to 21:00. Use the east entrance. 👩‍🔧 #museumlatenight A separate decorated version of the heading came from an old design file. The team must choose which wording to carry into its social text.

During review, the team also finds U+200B between two letters of entrance in the copied source. That code-point label is an editorial description, not text for visitors to read. The team confirms that the hidden space is accidental and removes only that occurrence in its revision. It does not apply the same removal to U+200D inside the technician emoji. The two invisible characters have different roles, so they need different decisions.

The revised notice reads: Late opening on Friday, 18:00 to 21:00. Use the east entrance. 👩‍🔧 #MuseumLateNight The editor chooses ordinary letters instead of the decorated heading and removes repeated emphasis. Hours, entrance, and intended tag wording remain. The technician emoji stays as optional decoration, with its complete sequence intact. These are explicit human choices; the tool has not restyled the heading, repaired the tag, or rewritten the message.

Compare both versions in the text checker, but do not approve the notice solely because findings decreased. The museum still needs a verified calendar date if Friday is ambiguous, confirmed entrance information, and a separate image review. It must check that any linked visitor page matches the notice. A useful review ends with a known draft, documented decisions, and specific remaining checks rather than an unsupported claim that the post is now accessible.

Test the final destination and keep an editable record

Copy the finished draft into its intended composer and inspect what survives. Check names, accents, emoji sequences, tags, addresses, and line breaks. If the destination changes the text or presentation, review that result as a new state. A plain-text comparison in one browser cannot establish what every app will display later. Keep the approved source available so the publisher can distinguish an intentional edit from an unexpected change introduced during transfer.

For an assistive technology review, use the relevant screen reader, language, and destination. Listen to essential instructions as well as unusual symbols. Note the environment and what you actually observed. A successful reading in one setup is useful evidence for that setup, not a universal result. When possible, include feedback from people who use the relevant technology in their ordinary work instead of assuming that a visual inspection answers the same question.

Check the rest of the post separately. An image may contain an entrance map, a video may introduce timing requirements, and text over media may need a contrast review. This tool cannot see or assess those elements and does not create their alternatives. The video safe-zone checker handles another limited visual preparation task; it does not turn this text review into a full media or accessibility audit.

Keep the original and revision when a character decision needs explanation. Download TXT for the actual text or explicitly save a JSON project to reopen the review; there is no autosave. A PDF report summarizes only the first 12 findings per selected draft, while CSV provides the full findings list. Treat private drafts and exported files carefully. Approval, native composition, and any later supported SMTasker account routine remain separate steps with their own owners.

Questions about Unicode & Accessible Social Text

Continue with another text review guide

Apply your edits in the Post Readability Checker, or explore another guide below. More useful browser-based helpers are in the free tools directory.