VIEW THIS AS

Auto mode follows the Route Engine until you choose a viewpoint.

YOU ARE HERE

ROUTE CHECK

CONNECTED TO

WHAT NEXT

Use the canonical route for this room, or HELP if you are unsure.

Top Ways to Translate Correctly | CLDR List Patterns, And/Or and Grammatical Joins

How do you translate a list correctly when English uses commas and ‘and’, French uses ‘et’, Spanish changes some conjunctions according to pronunciation, and a short mobile label has different spacing rules? The practical answer is to separate each list item from the joining pattern. Unicode CLDR list patterns provide locale-sensitive ways to connect two items, three items or longer sequences, with options for ‘and’, ‘or’, unit lists and shorter displays. This guide explains how to use them without losing any item, changing the logical relationship or mixing punctuation conventions.

A visibly wrong translation may contain perfect words but a broken list. ‘A, B, or C’ describes alternatives, while ‘A, B, and C’ may require all three. ‘Three feet, seven inches’ is a compound measurement, not a shopping list of independent objects. A target sentence can become misleading if translators type a literal English conjunction between already translated entries, add an extra comma, reorder values or cut a label to fit a button. A correct multilingual list keeps element identity and logical connection stable while formatting the join for the target audience.

Common searches answered here include how to translate ‘and’ and ‘or’ in lists, localize comma-separated lists, Unicode CLDR list formatting, internationalized list separators, Spanish conjunction y or e, list-formatting APIs, conjunction grammar, unit-list formatting and preserving list order in translation. The reader’s immediate rule is simple: classify the relationship, preserve the elements, select a target-locale pattern and verify the complete rendered sentence. Do not write one English-style joining rule into every language.

  • Separate list elements from the visible separators and conjunctions.
  • Never change ‘and’ into ‘or’ or reverse required and alternative membership.
  • Use locale-appropriate standard, or, short, narrow or unit patterns.
  • Test 0, 1, 2, 3 and longer lists, as well as nested or dynamic values.
  • Read grammatical context: inflection and pronunciation may depend on adjacent words.

First principles: why list joins are translation decisions

The list is data; the joining text is language

A reliable translation begins with an ordered collection of items and a separate statement about how those items relate. The locale formatter should not have to infer whether a list is exhaustive, optional or illustrative from punctuation. If a translated interface keeps items and connective language in one unstructured string, developers cannot safely insert or remove members at runtime.

To apply this principle, begin with the actual member objects and the purpose of the sentence rather than the surface punctuation. Compare a target-language rendering with the same input collection and record the changes attributable to formatting. Any loss of an element, altered choice requirement or modified value must be investigated as a content error, not waved away as a stylistic difference.

Conjunction is a logical operator as well as a word

A conjunction affects meaning. ‘Bring a pen and your ID’ makes a different request from ‘Bring a pen or your ID’. A translator must identify whether the source expresses inclusion, alternative, sequence or a compound measurement. A visually elegant separator cannot compensate for a changed requirement.

To apply this principle, begin with the actual member objects and the purpose of the sentence rather than the surface punctuation. Compare a target-language rendering with the same input collection and record the changes attributable to formatting. Any loss of an element, altered choice requirement or modified value must be investigated as a content error, not waved away as a stylistic difference.

Two, three and many items are different cases

The grammar of a two-item expression may differ from the end of a longer list. CLDR supplies patterns for two items and supports a special three-item form, with start, middle and end parts for longer sequences. Implementations should use the current locale data and an appropriate formatter rather than inserting a manually chosen word at the final position.

To apply this principle, begin with the actual member objects and the purpose of the sentence rather than the surface punctuation. Compare a target-language rendering with the same input collection and record the changes attributable to formatting. Any loss of an element, altered choice requirement or modified value must be investigated as a content error, not waved away as a stylistic difference.

List type selects the intended relationship

Standard, or and unit list types are not interchangeable. A shopping list, a set of acceptable submission formats, and a human-readable measurement need different joining rules. Short and narrow variants are also display choices, not licences to change meaning. Keep the type as an explicit parameter tied to the content model.

To apply this principle, begin with the actual member objects and the purpose of the sentence rather than the surface punctuation. Compare a target-language rendering with the same input collection and record the changes attributable to formatting. Any loss of an element, altered choice requirement or modified value must be investigated as a content error, not waved away as a stylistic difference.

Length affects presentation, not membership

A narrow label may omit some punctuation or use shorter separators in a supported locale. That should not silently drop a list member or convert ‘and’ to a different claim. When there is not enough room to display all options, use a deliberate summary with an accessible full list instead of truncating the serialized text.

To apply this principle, begin with the actual member objects and the purpose of the sentence rather than the surface punctuation. Compare a target-language rendering with the same input collection and record the changes attributable to formatting. Any loss of an element, altered choice requirement or modified value must be investigated as a content error, not waved away as a stylistic difference.

A listed phrase may need grammatical changes

List patterns handle joins, but a fully grammatical sentence may also require agreement, case or pronunciation-sensitive conjunction forms. Some language-specific rules involve the following word. The translator must check the complete output rather than assuming a generic punctuation library has solved every grammatical dependency.

To apply this principle, begin with the actual member objects and the purpose of the sentence rather than the surface punctuation. Compare a target-language rendering with the same input collection and record the changes attributable to formatting. Any loss of an element, altered choice requirement or modified value must be investigated as a content error, not waved away as a stylistic difference.

Unit lists are not itemized inventories

‘Three feet, seven inches’ and ‘two hours, five minutes’ describe parts of one quantity. They often use different spacing and punctuation from lists of food, contacts or choices. Preserve unit identities, values and ordering, and choose the unit-list family rather than formatting the output as ordinary English prose.

To apply this principle, begin with the actual member objects and the purpose of the sentence rather than the surface punctuation. Compare a target-language rendering with the same input collection and record the changes attributable to formatting. Any loss of an element, altered choice requirement or modified value must be investigated as a content error, not waved away as a stylistic difference.

Placeholder order protects identity

List patterns contain slots such as {0} and {1}. Inserting the wrong element into a slot changes who or what is listed. Translate the fixed joining text but not these placeholders; test that each input member appears exactly once unless a product-specific design intentionally says otherwise.

To apply this principle, begin with the actual member objects and the purpose of the sentence rather than the surface punctuation. Compare a target-language rendering with the same input collection and record the changes attributable to formatting. Any loss of an element, altered choice requirement or modified value must be investigated as a content error, not waved away as a stylistic difference.

Direction and script remain part of the rendering test

Names, numbers and right-to-left snippets can appear in the same list. Logical item order, display direction, spaces and punctuation require inspection in real browsers and assistive technologies. A plain-text screenshot in one language cannot establish that every mixed-script result reads correctly.

To apply this principle, begin with the actual member objects and the purpose of the sentence rather than the surface punctuation. Compare a target-language rendering with the same input collection and record the changes attributable to formatting. Any loss of an element, altered choice requirement or modified value must be investigated as a content error, not waved away as a stylistic difference.

The surrounding sentence sets the stakes

A list in a playful story can tolerate some stylistic flexibility. A medical instruction, eligibility rule or multiple-choice assessment may not. Preserve quantifiers and conditions surrounding the list—such as ‘at least one’, ‘all’, ‘except’, or ‘either’—because the connector alone does not encode the whole rule.

To apply this principle, begin with the actual member objects and the purpose of the sentence rather than the surface punctuation. Compare a target-language rendering with the same input collection and record the changes attributable to formatting. Any loss of an element, altered choice requirement or modified value must be investigated as a content error, not waved away as a stylistic difference.

A repeatable twelve-stage multilingual list workflow

Step 1: Find and delimit actual list items

Extract the ordered members from the source and identify which text belongs to each one. Do not split on commas blindly, because a person’s name or address may itself contain commas. Record the authoritative items and their boundaries before any translation.

The deliverable from this step should be explicit enough to test automatically. Record a sample input and an expected target-language output, including punctuation and order. Human review then focuses on the grammatical join rather than guessing what the data collection contained in the first place.

Step 2: Determine whether joining is conjunctive or alternative

Read the full sentence and annotate whether the list says all items apply, any single item can apply, or a special combination is intended. A translation should not substitute a target connective merely because it is more common in conversational writing.

A disciplined translation process distinguishes semantic facts from display decisions. The fact may be that three separate options exist, or that three separate pieces are mandatory. The chosen conjunction, spacing and punctuation express that fact in a locale. If the two layers disagree, correct the language layer without changing the underlying record.

Step 3: Distinguish ordinary lists from quantities

Determine whether the list describes choices, names, stages, dates or compound units. Where numeric units are involved, check whether the corresponding localization library supports unit-list formatting rather than normal comma conjunctions.

Perform a counterexample check before accepting the rule: imagine the same content with one fewer item, an embedded comma or a different script. If the output changes meaning, the method is too dependent on the original example. Document a boundary condition and add it to regression testing.

Step 4: Choose the target locale accurately

Identify the audience language and applicable region preferences. Do not assume the browser language always represents the content’s target locale; a user may read English content while running a different system interface.

Do not evaluate just the isolated phrase. Insert the formatted list into the surrounding sentence, display it in the final UI and ask a target-language reader to tell you what is required. A translation that sounds fluent in isolation may create the wrong instruction when combined with a quantifier or conditional.

Step 5: Choose a list pattern family

Select standard, or, unit, short or narrow deliberately. The family must match the semantic purpose of the data; screen width alone should not silently switch a list into alternatives or into a unit quantity.

The deliverable from this step should be explicit enough to test automatically. Record a sample input and an expected target-language output, including punctuation and order. Human review then focuses on the grammatical join rather than guessing what the data collection contained in the first place.

Step 6: Handle length and grammatical context

Use a formatting library or verified CLDR pattern to assemble the list of two or more items. Inspect the actual conjunction before words whose phonetics might trigger changes, and check whether the enclosing sentence requires a particular case or agreement.

A disciplined translation process distinguishes semantic facts from display decisions. The fact may be that three separate options exist, or that three separate pieces are mandatory. The chosen conjunction, spacing and punctuation express that fact in a locale. If the two layers disagree, correct the language layer without changing the underlying record.

Step 7: Keep input order and duplicates intentional

Some lists are ranked or procedural, while others are unordered sets. A translator should not alphabetize, reverse or deduplicate a source list without a documented semantic reason. Verify that the same intended entities survive translation.

Perform a counterexample check before accepting the rule: imagine the same content with one fewer item, an embedded comma or a different script. If the output changes meaning, the method is too dependent on the original example. Document a boundary condition and add it to regression testing.

Step 8: Protect dynamic placeholders

For labels such as ‘Alice, Bob and {count} others’, separate the dynamic count logic from the list elements. Do not translate interpolation tokens, field identifiers or code syntax; test a complete rendered string after interpolation.

Do not evaluate just the isolated phrase. Insert the formatted list into the surrounding sentence, display it in the final UI and ask a target-language reader to tell you what is required. A translation that sounds fluent in isolation may create the wrong instruction when combined with a quantifier or conditional.

Step 9: Test boundary sizes

Check zero, one, two, three, four and many entries. A formatter may produce an empty string for zero entries, but the surrounding sentence may need rewriting. Singular forms can also expose grammatical assumptions hidden by a longer list.

The deliverable from this step should be explicit enough to test automatically. Record a sample input and an expected target-language output, including punctuation and order. Human review then focuses on the grammatical join rather than guessing what the data collection contained in the first place.

Step 10: Test nested and mixed-script lists

Inspect lists where a single member contains commas, its own ‘and’, Arabic words, accented names or an embedded measurement. Use appropriate bidirectional rendering and avoid double-escaping punctuation.

A disciplined translation process distinguishes semantic facts from display decisions. The fact may be that three separate options exist, or that three separate pieces are mandatory. The chosen conjunction, spacing and punctuation express that fact in a locale. If the two layers disagree, correct the language layer without changing the underlying record.

Step 11: Compare the complete meaning against the source

Read source and target aloud with emphasis on all versus any. Verify each item appears once, the order remains justified, and a target reader can understand whether the list names requirements, options, examples or units.

Perform a counterexample check before accepting the rule: imagine the same content with one fewer item, an embedded comma or a different script. If the output changes meaning, the method is too dependent on the original example. Document a boundary condition and add it to regression testing.

Step 12: Document approved locale fallbacks and release tests

Track which library, locale data and rendering variants were used. Add regression cases for known tricky expressions and require a human review of high-risk output whenever a locale or formatting library is upgraded.

Do not evaluate just the isolated phrase. Insert the formatted list into the surrounding sentence, display it in the final UI and ask a target-language reader to tell you what is required. A translation that sounds fluent in isolation may create the wrong instruction when combined with a quantifier or conditional.

Sixteen worked examples: translation, joining and verification

Worked example 1: Three required documents

Visible problem: An application asks applicants to bring a passport, a signed form and a photo. A localization tool changes the final connector to a word meaning ‘or’.

Controlled translation decision: Keep all three requirements and format a conjunctive list. Confirm the requirement sentence contains no accidental permission to omit a document.

Verification: Compare a checklist of required items with the target rendering and ask whether a reader would bring the same materials.

The important test is whether the person reading only the target output would reconstruct the same membership and relationship as the source reader. Work from the underlying list entries, never from a comma-split string. That procedure makes it possible to improve fluency without changing what is present.

Worked example 2: Alternative file formats

Visible problem: A help page says upload a PNG, JPG or PDF; the target translation renders the items using an ‘and’ pattern.

Controlled translation decision: Use an alternative-list pattern, not a standard conjunction. Distinguish ‘any accepted format’ from an instruction to upload three separate files.

Verification: Test with three formats and a target-language speaker who has not read the source; their interpretation should remain one choice.

This example shows why a generic translation memory cannot own the whole solution. The members may be correctly translated yet the surrounding join may still be wrong. Ask which part is a language-specific separator, which part names an actual item, and whether another locale requires a different grammatical form.

Worked example 3: Two-item list

Visible problem: A site displays only two authorised pickup contacts and inserts a comma as though a third member were present.

Controlled translation decision: Select a two-item pattern that joins the names naturally, without an extra separator or missing conjunction.

Verification: Switch between two and three records to ensure no punctuation or membership is carried over.

A reliable release note for this case should name the input cardinality, the selected list pattern family, the target locale and the expected full rendered sentence. Reviewers can then reproduce the output after a library update or wording change instead of trusting a one-time screenshot.

Worked example 4: Three-item special form

Visible problem: A translator assumes the same two-item conjunction can be pasted onto a three-item list without inspecting punctuation.

Controlled translation decision: Use the locale-aware three-item behavior or start/end construction as defined by the formatter, not ad hoc English comma rules.

Verification: Compare the result with authentic target-language list usage, including its final separator.

If the list is part of a rule, document its logical effect before translation. Required items must remain required, alternatives must remain alternatives and compound measurements must remain one quantity. Aesthetic preference should not displace that invariant, even when the target sentence becomes slightly longer.

Worked example 5: Very long people list

Visible problem: A class directory displays thirty guardians using a single manually translated ‘and’, creating a confusing paragraph.

Controlled translation decision: Preserve the ordered records but consider an accessible list layout, truncation summary or pagination when presentation cannot support a readable sentence.

Verification: Check the complete data count and ensure a visual summary does not silently drop people or suggest exclusivity.

The important test is whether the person reading only the target output would reconstruct the same membership and relationship as the source reader. Work from the underlying list entries, never from a comma-split string. That procedure makes it possible to improve fluency without changing what is present.

Worked example 6: Nested list item with comma

Visible problem: A list includes ‘Paris, France’, ‘Rome, Italy’ and ‘Tokyo, Japan’, and a naive split on commas creates six city-country fragments.

Controlled translation decision: Store each city-country pair as one item, then apply a list formatter to the three structured items.

Verification: Inspect both the item count and the final punctuation; changing the locale must not split place names.

This example shows why a generic translation memory cannot own the whole solution. The members may be correctly translated yet the surrounding join may still be wrong. Ask which part is a language-specific separator, which part names an actual item, and whether another locale requires a different grammatical form.

Worked example 7: Spanish conjunction before a vowel sound

Visible problem: A Spanish target list includes conjunctions before words where y may conventionally change to e. A generic literal join may sound wrong.

Controlled translation decision: Use verified locale data and qualified linguistic review; consider phonetic context rather than a simplistic string replacement.

Verification: Check proper names, quoted words and numeral forms; a first-letter rule alone is not a reliable grammar engine.

A reliable release note for this case should name the input cardinality, the selected list pattern family, the target locale and the expected full rendered sentence. Reviewers can then reproduce the output after a library update or wording change instead of trusting a one-time screenshot.

Worked example 8: Spanish alternative before an o sound

Visible problem: A Spanish ‘or’ list has numbers and words whose pronunciation can affect use of o or u.

Controlled translation decision: Use language-aware formatting appropriate to the real spoken item, not a hardcoded replacement based on source spelling.

Verification: Test near-miss examples where typography and pronunciation diverge before approving an automatic pattern.

If the list is part of a rule, document its logical effect before translation. Required items must remain required, alternatives must remain alternatives and compound measurements must remain one quantity. Aesthetic preference should not displace that invariant, even when the target sentence becomes slightly longer.

Worked example 9: Hebrew mixed-script list

Visible problem: A Hebrew sentence joins a Hebrew item and a number written with Latin digits, and the punctuation appears in the wrong place.

Controlled translation decision: Apply the locale-aware join and handle bidirectional isolation for inserted items. Protect logical order independent of visual cursor behavior.

Verification: Examine actual browser output and screen-reader order, not just storage strings.

The important test is whether the person reading only the target output would reconstruct the same membership and relationship as the source reader. Work from the underlying list entries, never from a comma-split string. That procedure makes it possible to improve fluency without changing what is present.

Worked example 10: English Oxford comma preference

Visible problem: A brand team insists every language must use the serial comma conventions of its English style guide.

Controlled translation decision: Apply target-language punctuation conventions; source style preferences do not override legibility or the target locale’s normal grammar.

Verification: Check a three-item list in two target locales and ensure each uses appropriate punctuation without changing members.

This example shows why a generic translation memory cannot own the whole solution. The members may be correctly translated yet the surrounding join may still be wrong. Ask which part is a language-specific separator, which part names an actual item, and whether another locale requires a different grammatical form.

Worked example 11: Hours and minutes

Visible problem: An interface displays 2 hours, 5 minutes as a plain list of two independent entries with an awkward ‘and’.

Controlled translation decision: Use a unit or duration formatter that preserves number and unit relationships, and avoids a generic shopping-list pattern.

Verification: Compare the output with the underlying duration, including zero minutes and negative or approximate values when supported.

A reliable release note for this case should name the input cardinality, the selected list pattern family, the target locale and the expected full rendered sentence. Reviewers can then reproduce the output after a library update or wording change instead of trusting a one-time screenshot.

Worked example 12: Feet and inches

Visible problem: A measurement is supplied as three feet and seven inches, but translated using two ordinary object labels.

Controlled translation decision: Keep it as one compound measurement and use relevant unit-list formatting, without inventing a different dimension or conversion.

Verification: Verify numbers, unit abbreviations and the absence of an accidental ‘or’.

If the list is part of a rule, document its logical effect before translation. Required items must remain required, alternatives must remain alternatives and compound measurements must remain one quantity. Aesthetic preference should not displace that invariant, even when the target sentence becomes slightly longer.

Worked example 13: Screen width switches

Visible problem: A mobile button changes from standard to narrow list formatting and unexpectedly loses the final member.

Controlled translation decision: Use the documented narrow variant only to change joins and spacing; if content must be abbreviated, do that with an explicit accessible summary.

Verification: Resize the interface and compare the underlying list length across desktop and mobile.

The important test is whether the person reading only the target output would reconstruct the same membership and relationship as the source reader. Work from the underlying list entries, never from a comma-split string. That procedure makes it possible to improve fluency without changing what is present.

Worked example 14: Empty result set

Visible problem: A search widget says ‘Found: ‘ followed by an empty list, producing an unfinished target sentence.

Controlled translation decision: Handle zero results as its own localized message rather than passing empty items into grammar that expects a noun phrase.

Verification: Test zero and one items at the sentence level and confirm punctuation and number agreement.

This example shows why a generic translation memory cannot own the whole solution. The members may be correctly translated yet the surrounding join may still be wrong. Ask which part is a language-specific separator, which part names an actual item, and whether another locale requires a different grammatical form.

Worked example 15: Duplicate members

Visible problem: An application has two people both named Alex and a translator removes one as a presumed duplicate.

Controlled translation decision: Preserve distinct records even when human-readable labels match, unless there is an authorised data-level deduplication rule.

Verification: Check stable IDs, accessibility labels and the source count so that no person disappears.

A reliable release note for this case should name the input cardinality, the selected list pattern family, the target locale and the expected full rendered sentence. Reviewers can then reproduce the output after a library update or wording change instead of trusting a one-time screenshot.

Worked example 16: List inside a safety warning

Visible problem: A workshop warning requires gloves, a face shield and closed shoes. A translation uses a friendly short list but obscures that all are required.

Controlled translation decision: Treat the list as a safety requirement: preserve the conjunction and every protective item before optimizing typographic compactness.

Verification: Have an independent reviewer identify the mandatory items solely from the target-language notice.

If the list is part of a rule, document its logical effect before translation. Required items must remain required, alternatives must remain alternatives and compound measurements must remain one quantity. Aesthetic preference should not displace that invariant, even when the target sentence becomes slightly longer.

Failure modes: how to diagnose and repair them

1. Using English commas everywhere

Separators and conjunctions are locale data; a punctuation rule from one language can distort another. Diagnose the error by reconstructing the original array of items and the intended relationship, then comparing each element to the final target-language string. Repair the type selection, data boundary or locale-dependent join rather than merely replacing a visible punctuation mark. Retest at least two list lengths and inspect the complete surrounding sentence, because a local fix may create another error when the list grows or shrinks.

2. Switching and to or

This changes requirements into choices. Compare the logical function before polishing the prose. Diagnose the error by reconstructing the original array of items and the intended relationship, then comparing each element to the final target-language string. Repair the type selection, data boundary or locale-dependent join rather than merely replacing a visible punctuation mark. Retest at least two list lengths and inspect the complete surrounding sentence, because a local fix may create another error when the list grows or shrinks.

3. Splitting a member at an internal comma

An address or full name can contain punctuation. Use structured members, not raw text splitting. Diagnose the error by reconstructing the original array of items and the intended relationship, then comparing each element to the final target-language string. Repair the type selection, data boundary or locale-dependent join rather than merely replacing a visible punctuation mark. Retest at least two list lengths and inspect the complete surrounding sentence, because a local fix may create another error when the list grows or shrinks.

4. Reordering ranked lists

A ranked sequence is content, not decorative presentation. Translate each member without silently changing order. Diagnose the error by reconstructing the original array of items and the intended relationship, then comparing each element to the final target-language string. Repair the type selection, data boundary or locale-dependent join rather than merely replacing a visible punctuation mark. Retest at least two list lengths and inspect the complete surrounding sentence, because a local fix may create another error when the list grows or shrinks.

5. Dropping a duplicate-looking item

Two identical-looking labels can belong to different objects. Preserve IDs and intent. Diagnose the error by reconstructing the original array of items and the intended relationship, then comparing each element to the final target-language string. Repair the type selection, data boundary or locale-dependent join rather than merely replacing a visible punctuation mark. Retest at least two list lengths and inspect the complete surrounding sentence, because a local fix may create another error when the list grows or shrinks.

6. Formatting unit components as choices

Unit composition conveys one quantity, not alternatives. Use appropriate unit patterns. Diagnose the error by reconstructing the original array of items and the intended relationship, then comparing each element to the final target-language string. Repair the type selection, data boundary or locale-dependent join rather than merely replacing a visible punctuation mark. Retest at least two list lengths and inspect the complete surrounding sentence, because a local fix may create another error when the list grows or shrinks.

7. Hardcoding Spanish y/e rules by first letter

Pronunciation and special cases matter. Validate with reliable locale data and native review. Diagnose the error by reconstructing the original array of items and the intended relationship, then comparing each element to the final target-language string. Repair the type selection, data boundary or locale-dependent join rather than merely replacing a visible punctuation mark. Retest at least two list lengths and inspect the complete surrounding sentence, because a local fix may create another error when the list grows or shrinks.

8. Assuming short means fewer items

Short patterns change separators; substantive omission needs separate editorial approval. Diagnose the error by reconstructing the original array of items and the intended relationship, then comparing each element to the final target-language string. Repair the type selection, data boundary or locale-dependent join rather than merely replacing a visible punctuation mark. Retest at least two list lengths and inspect the complete surrounding sentence, because a local fix may create another error when the list grows or shrinks.

9. Breaking placeholders

Deleting {0} or repeating {1} can produce missing or duplicated list elements. Diagnose the error by reconstructing the original array of items and the intended relationship, then comparing each element to the final target-language string. Repair the type selection, data boundary or locale-dependent join rather than merely replacing a visible punctuation mark. Retest at least two list lengths and inspect the complete surrounding sentence, because a local fix may create another error when the list grows or shrinks.

10. Ignoring zero and one

An empty or singular list can require a different complete sentence, not a malformed join. Diagnose the error by reconstructing the original array of items and the intended relationship, then comparing each element to the final target-language string. Repair the type selection, data boundary or locale-dependent join rather than merely replacing a visible punctuation mark. Retest at least two list lengths and inspect the complete surrounding sentence, because a local fix may create another error when the list grows or shrinks.

11. Conflating display direction with order

Right-to-left presentation and logical membership order are different dimensions. Diagnose the error by reconstructing the original array of items and the intended relationship, then comparing each element to the final target-language string. Repair the type selection, data boundary or locale-dependent join rather than merely replacing a visible punctuation mark. Retest at least two list lengths and inspect the complete surrounding sentence, because a local fix may create another error when the list grows or shrinks.

12. Trusting a fluent target list without checking logic

A native-sounding list can still express the opposite obligation or omit a crucial item. Diagnose the error by reconstructing the original array of items and the intended relationship, then comparing each element to the final target-language string. Repair the type selection, data boundary or locale-dependent join rather than merely replacing a visible punctuation mark. Retest at least two list lengths and inspect the complete surrounding sentence, because a local fix may create another error when the list grows or shrinks.

Practice and transfer with model reasoning

Practice 1: A form says choose JPG and PDF although users can upload either type.

First identify the members, choose whether their relationship is conjunctive, alternative or unit composition, and explain what may be localized. Write the expected complete message before reading the model route.

Model route: Choose an or-pattern and confirm the help text names two accepted alternatives. After implementing that decision, test the list with a different number of items and a target-language reader who has not seen the source. Their explanation of the relationship is your functional quality check.

Practice 2: A laboratory notice requires gloves, mask and shoes.

First identify the members, choose whether their relationship is conjunctive, alternative or unit composition, and explain what may be localized. Write the expected complete message before reading the model route.

Model route: Use a conjunctive list and verify all protective items remain mandatory. After implementing that decision, test the list with a different number of items and a target-language reader who has not seen the source. Their explanation of the relationship is your functional quality check.

Practice 3: An item is ‘Paris, France’ inside a three-location list.

First identify the members, choose whether their relationship is conjunctive, alternative or unit composition, and explain what may be localized. Write the expected complete message before reading the model route.

Model route: Keep the comma-containing phrase as one structured item. After implementing that decision, test the list with a different number of items and a target-language reader who has not seen the source. Their explanation of the relationship is your functional quality check.

Practice 4: A dynamic list changes from three members to one.

First identify the members, choose whether their relationship is conjunctive, alternative or unit composition, and explain what may be localized. Write the expected complete message before reading the model route.

Model route: Render the single member without a dangling connector or preceding comma. After implementing that decision, test the list with a different number of items and a target-language reader who has not seen the source. Their explanation of the relationship is your functional quality check.

Practice 5: A Spanish list has a conjunction before an expression with a special sound.

First identify the members, choose whether their relationship is conjunctive, alternative or unit composition, and explain what may be localized. Write the expected complete message before reading the model route.

Model route: Check the relevant grammatical pattern and pronunciation; do not replace mechanically based on its first written letter. After implementing that decision, test the list with a different number of items and a target-language reader who has not seen the source. Their explanation of the relationship is your functional quality check.

Practice 6: A mobile interface hides two members from a narrow joined list.

First identify the members, choose whether their relationship is conjunctive, alternative or unit composition, and explain what may be localized. Write the expected complete message before reading the model route.

Model route: Provide an explicit summary and accessible full membership rather than calling the result an ordinary narrow list pattern. After implementing that decision, test the list with a different number of items and a target-language reader who has not seen the source. Their explanation of the relationship is your functional quality check.

Practice 7: A target text joins a two-hour five-minute duration as options.

First identify the members, choose whether their relationship is conjunctive, alternative or unit composition, and explain what may be localized. Write the expected complete message before reading the model route.

Model route: Use unit-duration formatting and preserve the one quantity. After implementing that decision, test the list with a different number of items and a target-language reader who has not seen the source. Their explanation of the relationship is your functional quality check.

Practice 8: Two records share the same display label.

First identify the members, choose whether their relationship is conjunctive, alternative or unit composition, and explain what may be localized. Write the expected complete message before reading the model route.

Model route: Keep both records identifiable with stable IDs and appropriate context. After implementing that decision, test the list with a different number of items and a target-language reader who has not seen the source. Their explanation of the relationship is your functional quality check.

Practice 9: A Hebrew list contains an English product code.

First identify the members, choose whether their relationship is conjunctive, alternative or unit composition, and explain what may be localized. Write the expected complete message before reading the model route.

Model route: Test logical item order, bidi display and punctuation around the code. After implementing that decision, test the list with a different number of items and a target-language reader who has not seen the source. Their explanation of the relationship is your functional quality check.

Practice 10: An English brand style guide requires serial commas worldwide.

First identify the members, choose whether their relationship is conjunctive, alternative or unit composition, and explain what may be localized. Write the expected complete message before reading the model route.

Model route: Follow the target locale’s legitimate conventions while retaining the same semantic list. After implementing that decision, test the list with a different number of items and a target-language reader who has not seen the source. Their explanation of the relationship is your functional quality check.

Practice 11: A translation deletes the {1} slot in a list pattern.

First identify the members, choose whether their relationship is conjunctive, alternative or unit composition, and explain what may be localized. Write the expected complete message before reading the model route.

Model route: Restore the formatter-controlled placeholder and verify no input member is lost. After implementing that decision, test the list with a different number of items and a target-language reader who has not seen the source. Their explanation of the relationship is your functional quality check.

Practice 12: A choice list contains no available options.

First identify the members, choose whether their relationship is conjunctive, alternative or unit composition, and explain what may be localized. Write the expected complete message before reading the model route.

Model route: Use a complete zero-choice message and avoid an unfinished sentence. After implementing that decision, test the list with a different number of items and a target-language reader who has not seen the source. Their explanation of the relationship is your functional quality check.

Frequently asked questions

What is a CLDR list pattern?

It is locale data describing how to join an ordered collection into readable text, including patterns for two items and for the start, middle and end of longer lists. The formatter applies the selected list type to structured input. For quality assurance, inspect both the structured input and the final string. A multilingual list succeeds when readers understand the same people, objects, alternatives or requirements in the target language without relying on the English punctuation that happened to appear in the original.

Is the Oxford comma required in every language?

No. List punctuation differs by locale, style and context. Preserve meaning and use verified target conventions rather than imposing one English editorial rule globally. For quality assurance, inspect both the structured input and the final string. A multilingual list succeeds when readers understand the same people, objects, alternatives or requirements in the target language without relying on the English punctuation that happened to appear in the original.

Does a list formatter translate the list items?

No. It handles joining text and separators. Each element still needs correct translation, naming, unit formatting and contextual grammar before insertion. For quality assurance, inspect both the structured input and the final string. A multilingual list succeeds when readers understand the same people, objects, alternatives or requirements in the target language without relying on the English punctuation that happened to appear in the original.

What is the difference between standard and or patterns?

Standard joins a conjunctive list; or indicates alternatives. The distinction can affect compliance, eligibility and safety. For quality assurance, inspect both the structured input and the final string. A multilingual list succeeds when readers understand the same people, objects, alternatives or requirements in the target language without relying on the English punctuation that happened to appear in the original.

Are short and narrow variants safe?

They are safe when used for the intended display type and supported locale, with complete membership retained. They do not authorize dropping items. For quality assurance, inspect both the structured input and the final string. A multilingual list succeeds when readers understand the same people, objects, alternatives or requirements in the target language without relying on the English punctuation that happened to appear in the original.

How do list patterns handle three or more items?

The relevant locale data may provide a direct three-item pattern or a sequence of start, middle and end parts that combine all elements in order. For quality assurance, inspect both the structured input and the final string. A multilingual list succeeds when readers understand the same people, objects, alternatives or requirements in the target language without relying on the English punctuation that happened to appear in the original.

Can I split a string on commas and use list formatting?

That is unsafe if an element itself contains commas, quotes or special punctuation. Prefer an actual array of items. For quality assurance, inspect both the structured input and the final string. A multilingual list succeeds when readers understand the same people, objects, alternatives or requirements in the target language without relying on the English punctuation that happened to appear in the original.

Why do some conjunctions change before another word?

Pronunciation-sensitive grammatical rules can affect conjunction choices in certain languages. A faithful translation needs authentic target-language behavior, not just fixed character substitutions. For quality assurance, inspect both the structured input and the final string. A multilingual list succeeds when readers understand the same people, objects, alternatives or requirements in the target language without relying on the English punctuation that happened to appear in the original.

What about right-to-left languages?

Use locale-sensitive joins and proper handling of embedded left-to-right values such as numbers or product codes. Test both visual display and reading order. For quality assurance, inspect both the structured input and the final string. A multilingual list succeeds when readers understand the same people, objects, alternatives or requirements in the target language without relying on the English punctuation that happened to appear in the original.

How should I test an internationalized list?

Use realistic examples containing 0, 1, 2, 3 and many items, nested punctuation, alternative versus required membership, mixed scripts and mobile layouts, then have target-language users verify meaning. For quality assurance, inspect both the structured input and the final string. A multilingual list succeeds when readers understand the same people, objects, alternatives or requirements in the target language without relying on the English punctuation that happened to appear in the original.

A practical translation QA checklist

  • Original list items were extracted as structured data, including embedded commas and names.
  • List type is recorded as standard, or, unit or a specified supported variant.
  • Cardinality tests include zero, one, two, three and many items.
  • Every output contains exactly the authorised members in the intended order.
  • Pronunciation-sensitive conjunctions and right-to-left rendering receive target-language checks.
  • The whole sentence preserves all/any logic, qualifications, units and conditions.
  • Translated display text, IDs, placeholders and measurements were reviewed separately.

Contextual reading routes and sources

The most transferable lesson is to protect the collection and its logical relationship first, then let the target language express that relationship naturally. A well-localized sentence changes punctuation and joining words where appropriate, but it never asks a reader to bring different documents, choose different options, or interpret a different quantity.