A name one way, a code the other: two text directions in an interface
Test mixed text directions through entry, display, selection and copying while preserving original names and identifiers.
Article language: English

In this article
A customer's name looks correct until a project code appears next to it. Suddenly a bracket is on the other side, the plus sign has separated from a phone number and an employee starts copying the value by hand. A site does not need an Arabic language version to encounter text read from right to left. A form into which someone enters their name is enough.
The problem concerns the relationship between characters, not just the choice of script. A Polish field label, a name in another alphabet, a Latin identifier and a number can appear on one line. Each part has its own meaning, and the user should be able to read and copy the right value. A design cannot expect them to repair the order by adding spaces themselves.
A line contains several pieces of information
Imagine the fictional dashboard of Rama, a studio handling orders from different exhibition organisers. The dashboard is in Polish. In one request, the organisation name is written in Arabic script, the project code is RM-208-A, and the contact has supplied a phone number in international format. All three values must be preserved and presented so they can be recognised.
First establish the structure: the organisation name is a name, the code identifies the project and the number leads to a contact. Do not concatenate them into one unexplained fragment just because it fits on a wide mockup. Separate labels or clear groups can help people recognise the values even when they cannot read that alphabet. They do not replace correct handling of text direction, though.
Keep each label associated with its value when the layout changes. If a name moves above the code on a phone, it must not look like the code's description or part of the next item's heading. With several requests, clear card boundaries help prevent a number being assigned to the neighbouring organisation. A direction change should not weaken ordinary information hierarchy.
The Unicode Bidirectional Algorithm, UAX #9 distinguishes the logical character order of stored text from presentation order. The version read on 30 October 2026 was revision 52 for Unicode 18.0.0, dated 1 September 2026. The state of the specification does not establish identical handling in every environment in use.
For Rama this means two checks. One examines how information looks in the dashboard. The other compares the value retained and passed on. Comparing values requires text, not just a screenshot. An image shows presentation but cannot establish the field contents after copying. Keep an approved sample and check it at subsequent points in the flow.
Do not manually reverse letters to repair one screen. That may make a view look familiar to someone who cannot read the text, while the copied content no longer matches the original.
Do not confuse direction with alignment either. Moving the whole field to the right edge does not settle the relationship between the name, bracket and number within the value. Likewise, the page language does not tell you the direction of every name a user supplies. Describe requirements for the data the application actually accepts.
When launching language versions of a site (in Polish) (article in Polish), you check the consistency of the whole journey among other things. Fragments in the opposite writing direction also need acceptance testing in the Polish version. Do not put that off until another market launches. A name can appear in a form before an expansion plan does.
The boundary of a name should not disappear when a number is added
A request heading shows the organisation name, a hyphen and the number of attachments. The name is an independent value; the count describes something else. If the design treats them as a single string without distinguishing its parts, some combinations of characters may produce a misleading layout. Acceptance testing should cover the whole combination, not just the name in an empty field.
W3C guidance on mixing directions in HTML describes wrapping a phrase in appropriate markup with a dir direction, and, for unknown direction, options including dir="auto" or the bdi element. The aim is to define the fragment's boundary and direction. The implementation determines the mechanism; merely moving text on a mockup does not fulfil the guidance.
Show which data forms one fragment and which values remain separate. A file count should remain recognisably distinct from the customer name. The project code should be recognisable together with its hyphens, not as loose digits appended to a name. Give that structure and the sample data to the implementer, rather than only a screenshot of a correct case.
For a user-supplied name, do not assume a fixed direction from one record. The next request may contain a Latin or mixed name, or one beginning with digits. W3C describes automatic direction based on the first strong directional character. That does not mean the language or intent of every complex name has been guessed; difficult examples still need acceptance testing.
Prepare several combinations:
- A name in another direction next to a Polish label and a request number.
- A name also containing a Latin part, a digit or punctuation.
- A code with hyphens or brackets, shown both separately and next to a name.
- A phone number with a plus sign and spaces, shown as text and as a contact action.
- A short list of two names separated by a comma, with a separate result count.
Each case should have an agreed source value and an expected way of reading it. A person who knows the relevant language confirms the names and their meaning. The designer and implementer check the structure, presentation and interface behaviour. This is collaboration in acceptance testing, not a substitute for language expertise through visual intuition.
Also prepare a reference for someone who does not know the alphabet but handles the request. They should recognise the labels, code and correct action without rewriting the name their own way. Do not require them to transliterate it, meaning write it in another alphabet, if the process must preserve the original value. Agree the source of any additional helper name separately.

Do not assume one correct case settles all the others. Changing a label, adding a bracket or adding a number can reveal a problem absent when the name stands alone. Documentation examples show a mechanism; your own trial should cover the actual structure of the headings, search results and confirmations used by Rama.
Test a longer name after it wraps onto another line. The article on long names and text in interfaces (in Polish) (in Polish) helps preserve the necessary scope, but here the relationship between directions adds another concern. Truncating the end may remove the part that distinguishes two organisations. The user should have an agreed way to access the full value.
Also check the code itself after truncation. RM-208-A and RM-208-B differ at the end; a preview showing only the beginning hides the decisive part. Expanding the full value should work without hovering a mouse. Selection or copy buttons should act on the whole identifier, not only the characters left in a short label.
Entry, selection and copying are separate moments
A form accepts a name, shows it in a summary and then in a search result. Each view may have a different surrounding structure. Do not stop testing at the edit field. Follow the whole path to the point from which an employee prepares a response or hands the case on.
In the field, test typing and pasting the approved sample. The user should be able to select part, correct it and continue. If the interface offers suggestions, check that the complete name and code can be recognised. Choosing a specific record from an autocomplete field (in Polish) (article in Polish) cannot be replaced by visible text merely looking similar.
Test the error message too. If a name fails validation, the user should learn the actual reason and an allowed way to correct it. A generic “use valid characters” may suggest the alphabet itself is forbidden when another field rule is the problem. Decide deliberately what data to accept and check it against how the data will later be stored.
Separate appearance testing from copy testing. An employee copies RM-208-A into a controlled comparison field. The tester checks that the correct value and character order survived. Then they compare a copy of the whole line: name, code and additional information. The two actions may include different amounts of text, which the button label must make clear.
“Copy code” should copy the code, not the entire card layout. “Copy contact details” includes a different set. If an action adds labels or strips formatting, decide that and show it in the sample. Acceptance testing should compare the result at the destination rather than stop at a green tick beside the button.
After pasting a name into another form, reread the saved request. A correct clipboard does not prove the next application preserved the value. If the process involves a manual correction, record why and check that it does not fix only the appearance at the expense of the data. You can then identify where the value changed rather than repeatedly fixing every view.
For a phone number, also compare the visible value with the destination of the contact action. An attractive display does not prove the button uses the right number. Prepare controlled data and test conditions that do not place a call to a random person. The design should distinguish copying a value from initiating contact.
If the organisation uses spoken output or generated messages, separately test pronunciation of names and units (in Polish) (article in Polish). Correct order on screen does not establish correct pronunciation. Keep the tests distinct so a good result in one does not conceal the absence of the other.
Aeneia Design
For forms and results containing different text directions, prepare samples for the whole path: entry, display, selection and copying. Aeneia Design can help organise the interface structure and acceptance of its behaviour using approved language material. Prepare an interface for names in different alphabets
Check the data beyond the wide screen
On a phone, a heading occupies more lines, buttons move below the name and the code may be truncated. Do not assess this solely by shrinking a mockup. In a working version, check wrapping, the full view and copying. Choose the actual browsers and devices included in the deployment scope instead of claiming support for every possible environment.
A file export and confirmation message need separate checks. A document may compose text using a different mechanism from the website. Decide which destinations belong to the delivery scope and who is responsible for accepting them. In the report, separate the dashboard result from the generated document result even when both use the same data.
Keep the same control set for each destination: the source value, a description of its surroundings and the expected copy scope. After a fix, repeat the specific case and the other locations using the changed component. Once a heading is fixed, check the number in the summary and the entry in search results again. Record environments that were not tested too.
If names go through AI or translation, retain the source material for comparison. Checking document translation (in Polish) (article in Polish) also covers consistency of names and identifiers. An appearance error does not necessarily mean the stored value changed, and a real change in value will not always be visible in a short preview. Both questions need answers.
Requirements for Aeneia Design can be described without a ready-made technical recipe: this name, this code and this number must remain recognisable in the form, summary and copy. Approved samples and a description of the goal establish a testable scope. The employee should be able to handle the request. Rearranging brackets should not become an extra speciality.







