Gains:
- Ability to distinguish localization from translation and manage placeholder integrity, text length and date/money/measure format correctly
- Ability to adapt technical features such as plural rules and right-to-left languages to the target locale
- Ability to evaluate cultural elements from the perspective of a local expert and capture the placeholder and cultural risks of artificial intelligence with QA
Turning an app's "Save" button to "Save" is a translation; But adapting that application's date format, currency, right-to-left writing, button length, cultural images and legal texts to the target market is localization. In this unit, you will learn about localization, its technical features (placeholder, length, coding), the role of AI in this process, and cultural adaptation. The goal is to think like a localization expert, “making the product fit the target culture, not the words.”
Basic concepts
Localization (L10n — localization; L10n because there are 10 letters between “l” and “n”) is the process of fully adapting a product (software, web, game, application) to a specific language and culture; It includes but transcends translation. Internationalization (i18n — internationalization) is the act of designing the product from the ground up to be very language-ready (separating text from code, allowing length flexibility); It precedes and enables localization.
String is the piece of text to be translated in the software. Placeholders are marks within the string that are populated with the variable at runtime: "Hello {name}", "{count} items". Locale is the combination of language + region (tr-TR, en-US); Specifies the date, time, number and currency format.
Localization differs from translation: you convey not just the meaning, but also function and cultural appropriateness. A "3/4/2026" date is March 4 in the USA and meaningless in Türkiye (we write 03.4.2026); "₺" instead of "$"; The color red may be a warning in one culture and a celebration in another.
Tip: What translators most often miss in localization are elements outside the text: date/time/number format, currency, unit of measurement (miles/km), first-name order, address format, phone format. Scan for these on every project with a "locale checklist".
Placeholder and technical integrity
The most dangerous technical mistake in localization is corrupting placeholders and tags. If you delete the {n} in the sentence "You have {n} messages", write it wrong, or put it in the wrong place according to Turkish syntax, the software will crash or appear crude as "You have {n} messages". Rules:
- Never rotate, delete, or format placeholders. {name}, %s, {{count}} remain the same.
- Turkish syntax can replace the placeholder; Move it to the new location, preserving the meaning, but do not destroy the sign itself.
- Plural rules vary depending on the language: While English says "1 item / 2 items", in Turkish there is no plural suffix after the number ("2 items"). Localization frameworks handle this separately.
The AI is a two-pronged tool here: it translates strings quickly, but can accidentally flip or lose the placeholder. That's why a placeholder QA round is essential in localization.
Caution: Text expansion is the hidden problem of localization. The translation text from English to Turkish is often 20-40% longer; "OK" is 2 letters, its counterpart "OK" is 5 letters. Translation that doesn't fit into a narrow button breaks the interface. If possible, see in the real interface if the target text fits.
Localization flow and cultural adaptation with AI
AI speeds up the following tasks in localization: initial translation of strings, consistency checking, length warning (“this translation is 35% longer than the original”), cultural appropriateness screening (“will this image/example cause problems in the target culture?”). But the cultural decision belongs to the human: the local expert knows how a joke, a holiday, an example, a color will be perceived in the target culture. AI can give a general warning; The final decision is made by the translator who knows the local market.
Examples of cultural adaptations: payment methods (local cards), example names (local names), units of measurement, legal obligations (KVKK/GDPR texts), holidays, form of address (you/you), color and symbol meanings.
three mini cases
Case 1 — Placeholder QA prevented the crash. In a mobile application's translation of 1,200 strings, the AI translated the {count} placeholder as "{number}" in 18 places. The Placeholder QA round caught these; If it wasn't fixed, the application would crash on those screens.
Case 2 — Length broke the interface. One software's menu was designed in English; When the Turkish translations became 30% longer on average, three menu items were moved and cut. If the team had received the length warning early, they would have prepared short alternatives (abbreviation if necessary, instead of "Settings"); the job was redone and updated with process length control.
Case 3 — Cultural adaptation saved the sale. In a game promotion there was a merit badge with a pig figure; in a target market this was culturally inappropriate. The local translator warned, the figure was changed. The AI had translated the text, but it was the local expert who pointed out the cultural risk.
Four copyable templates
1) String translation (placeholder protected):
Translate the following software strings into [target language]. RULE: NEVER translate, delete, or format placeholders such as {name}, %s, {{count}}; keep it as it is (you can move it according to Turkish syntax). Preserve HTML/tags. Write it concisely and suitable for the interface. Format: source → translation.Strings: [...]
2) Placeholder/label QA:
Below are the source and translated strings. Flag placeholder and tag issues ONLY: translated/deleted/corrupted{...}, %s, {{...}}, <tag>. List how many placeholders are in the source, how many are in the translation, and list the ones that do not match. Source: [...] | Translation: [...]
3) Length and interface warning:
Evaluate the following UI translations for length. For each translation, give the percentage extension according to the source and mark those that may not fit into tight spaces (buttons, menus). For those that don't fit, suggest a shorter alternative that preserves the meaning. Pairs (source | translation): [...]
4) Cultural suitability screening:
Your role: [target market] localization consultant. Flag elements in the following content that may cause problems in the target culture: image, example, name, color, symbol, joke, date/measurement format, legal text. The final decision is mine; you point out the risk and suggest alternatives.Content: [...]
Weak prompt / Strong prompt
Weak: "Translate these app texts." (Placeholder, length, no interface context; machine translates placeholder, text gets longer.)
Strong: "Translate these mobile application strings to Turkish. Keep the {user} and %d placeholders as they are. These texts will appear in narrow buttons; keep them short if possible. 'Settings'→'Settings', 'Profile'→'Profil'. Follow the Turkish rule for plural expressions (no plural suffix after the number)."
Difference: strong prompt gives placeholder, length, term and plural convention; the output would be close to entering the interface directly.
Localization dimensions table
Size
example
Risk
Placeholder/label
{name}, %s, <b>
Software crashes
length
"OK"→"OK" (150%)
Interface overflows
Date/number/money
3/4/26, $, 1,000.50
false information
plural rule
2 items → 2 items
bad grammar
cultural element
image, color, humor
reputation/sales
legal text
KVKK/GDPR
legal risk
Common mistakes
- Flip/delete placeholder. It causes the software to crash or raw text to appear.
- Not taking into account text stretching. Interface overflows, elements are truncated.
- Not converting date/currency/measurement format. "5 miles" remained, not "8 km".
- Passing the cultural element without consulting the local expert. Reputation and sales risk.
- Translating plural rules with English logic. Bad grammar like "2 items".
Pseudo-localization and right-to-left languages
Two technical issues determine localization quality. The first is pseudo-localization: testing the product with fake but realistic text of length and special characters (e.g. "Settings" → "[Ŝéttîngŝ~~]") before the actual translation. This shows whether the interface can handle long texts and special characters, whether strings are actually being extracted before the translation starts. If the translator working with the developer recommends this test, many interface errors will be caught before they occur.
The second is right-to-left (RTL) languages: Languages like Arabic, Hebrew, Persian are written right-to-left, and localization requires mirroring not just the text but the entire interface layout (menu position, arrows, alignment). In an RTL translation, numbers and Latin letter terms can create confusion; This problem of "bidi text" requires special attention. AI can translate RTL text, but layout mirroring and two-way flow decisions require technical-cultural expertise. These two issues show that localization is an engineering-cultural work beyond translation.
In summary
Localization means adapting the product, not the words, to the target language and culture; includes translation, but also includes placeholder integrity, text length, date/money/measure format, plural rules, and cultural elements. AI accelerates string translation, length and cultural risk screening; But a QA tour is essential as it can disrupt the placeholder and the cultural decision is made by the expert who knows the local market. Success in localization is attention to detail beyond the text.
Application task
Take a sample interface text of 15-20 strings (with placeholders {...} or %s and a date/money example). Translate with the "String translation" pattern, then check placeholder integrity with "placeholder QA" and check overflow risk with "length warning". Adapt the date and money format to the target locale and perform a "cultural suitability scan" if there is a cultural element.
checklist
- [ ] I kept the placeholder and labels as they were and confirmed them with QA.
- [ ] I controlled text stretching and prevented overflow in narrow areas.
- [ ] I adapted the date, number, currency and measurement units to the target locale.
- [ ] I translated plural expressions according to the rules of the target language.
- [ ] I evaluated the cultural elements from the perspective of a local expert.