Icon Localisation: Broken Metaphors and the Icon Problem

Say localisation and text comes to mind first. Yet the first thing a user sees in an interface is often not text but an icon. Icons look as though they were mute, and for that reason they are left outside the scope of translation. In reality icons are every bit as cultural as words; they take their meaning from the context they were born into. Icon localisation is therefore a heading that goes undiscussed in most projects and yet bears directly on the outcome.

Contents

The book calls this area the visual-iconic level and counts it among the fundamental dimensions subject to cultural adaptation. Culture-specific icons that are not understood in the target market are a classic localisation problem (Jiménez-Crespo 2024, p. 101). The problem generally stays invisible, because nobody reports that they have failed to understand an icon.

In this article we look at how icon localisation works. We take up the notion of the broken metaphor, the classic examples given in the book, why the use of flags is held to be problematic, and the path that can be followed in reaching a decision. The aim is to move icons out of the category of design decision and into that of localisation decision.

What is the visual-iconic level?

The classification Tercedor proposes for software localisation counts five dimensions open to cultural adaptation. One of them is the visual-iconic level, which deals directly with icons that can be understood in the target locale (Jiménez-Crespo 2024, p. 101). The technical level alongside it covers colour, pictogram and layout preferences.

The power of icons comes from their recognisability. The moment an icon is recognised there is no need to read, and that lowers the cognitive load. But recognisability is learned, not innate. An icon that one community of users has grown used to over the years may set off no association at all in another.

Icon decisions therefore ask two questions at once. First, is the object the icon depicts known in the target culture? Second, is the link between that object and the function intended made in the target language? The two questions differ, and most errors arise from the second. Even where the object is recognised, if the link between it and the function is not made in the target language the icon becomes mere decoration.

The broken metaphor: does a house icon mean home page?

The most illuminating example the book gives is the house icon. Many browsers, Internet Explorer among them, used a house icon for the “Home” function. In English the link between the word and the icon is a natural one. In Spanish the same function is localised as “Página principal”, that is, main page; in French it becomes “accueil”, meaning welcome (Jiménez-Crespo 2024, p. 101).

This is what is called a broken metaphor. The relationship between the visual icon and the language of development is meaningful in the source language; after localisation the analogy disappears in some locales (Jiménez-Crespo 2024, p. 109). The icon stays where it is, but it no longer calls anything to mind.

The same disconnect holds for Turkish. The phrase “ana sayfa” carries no idea of a house in it. Users have learned the house icon by heart over the years and have no trouble with it today; that does not mean the metaphor works, though. It only means that habit has taken the metaphor’s place. A newly designed icon enjoys no such indulgence; it has to be understood at first glance.

Classic examples: the briefcase and the mailbox

Another example mentioned in the book is the “My Briefcase” icon in older versions of Windows. It was used for taking files home on an external drive, but users could not work out what it meant (Jiménez-Crespo 2024, p. 101). The icon presumed a working habit, and that habit was not universal.

The examples Dietz reports point to the same problem. Many visual metaphors specific to American culture are unintelligible in other countries: using the home plate on a baseball field for the home page, say, or showing a typical American mailbox with a red flag to signal that a new message has arrived (Jiménez-Crespo 2024, p. 103).

Objects that grow obsolete over time are added to the same list. The book also counts the tray-shaped icons used for composing a new message or reaching the inbox among the examples that cause confusion on a global scale. Most of today’s users have never seen such trays in real life.

Why is using flags problematic?

Using flags in language selection menus is a widespread habit. The book relays the argument that finds it culturally inappropriate: using a national flag to indicate the version of a language equates the language with the country (Jiménez-Crespo 2024, p. 109). Language and country, however, are not the same thing.

The problem does not stay theoretical. Choosing a single country’s flag for Arabic shuts out every other speaker. The same holds for Spanish, Portuguese, French and English. When users do not see their own flag, they conclude that the product is not aimed at them.

The solution in practice is simple. In the language selection list, each language’s own name is written in that language’s own spelling: Türkçe, Deutsch, Español. Where a country-specific locale is needed, language and country are given together. A flag is meaningful only where a country is actually being chosen, in a delivery country list for instance.

The same logic applies to other visual shortcuts. Representing a region with a single building, a single dish or a single item of clothing is a reduction of the same kind. The book points out that choices of this sort can create the impression that the product is not speaking to that user. Visual shortcuts save time, but who they leave out has to be asked every single time.

Four options in icon localisation

When you notice that an icon is not working in the target market, there is no single remedy. There are four routes, and each carries a different cost. The table below compares those routes, when each one is preferable and what its side effects are.

Option	When is it suitable?	Side effect
Keep the icon	Recognisability is high, the habit is established	May stay meaningless for a new user
Add a text label	The link of meaning has snapped but the icon is familiar	Takes up interface space, lengthens the translation
Change the icon	The object depicted is not known locally	Inconsistency between versions
Move to an abstract icon	The same function will be used in many markets	Longer learning time

In most situations the second option is the most sensible. Keeping the icon secures consistency across versions, while the text label re-establishes the link of meaning. The combination brings a gain in accessibility as well; for users of screen readers a label is mandatory in any case.

How do you test icons?

Icon decisions have to be reached by testing rather than by debate. One simple method runs like this: you show a small group of target users the icons alone and ask what they are for. You hide the labels. If the rate of correct guesses is low, that icon is not working in the target market.

The second method is testing with a real task. You ask the user to complete a particular job and watch which icon they reach in how many seconds. Hesitation time brings icons that are not understood to light more reliably than survey questions do.

Third, the support records need looking at. Reports from users along the lines of “I can’t find that button” are in fact reports of an icon problem. These records generally never reach the translation team, and yet they are the cheapest source of feedback on localisation quality. Even a short quarterly meeting between the support team and the translation team makes a difference.

Beyond icons: navigation metaphors

Cognitive aspects are counted among the dimensions of cultural adaptation as well: navigation, metaphors, mental models and modes of interaction (Jiménez-Crespo 2024, p. 102). This is a deeper layer than individual icons. How a menu is constructed, in what order information is presented and how the user predicts the next step all belong to it.

The desktop metaphor is the best-known example of it. The notions of folder, file, wastebasket and desktop are taken from a particular office culture. Today these notions have been learned globally, but their being learned does not make them natural. When designing a new metaphor, the same assumption should not be repeated.

The practical upshot is this: changing an icon sometimes treats the symptom. If a user cannot find a function, the problem may lie not in the icon but in the way the information has been grouped. That is why it is valuable for localisation reports not to stop at the level of words but to report blockages at the level of navigation too.

The Türkiye context

The break met with most often in Turkish is in the shopping icon. The wheeled supermarket trolley that is common in English interfaces is named with the word “sepet”, basket, in Turkish. The icon shows a trolley, the label says basket. Users have grown used to it, but the two elements do not support one another; for a new user the value of the cue is low.

The question of flags has a dimension of its own where Türkiye is concerned. Turkish is not spoken only in Türkiye; there is a substantial Turkish-speaking user base in Germany, the Netherlands, Cyprus and many other countries. Using a flag in language selection ignores that audience. The simple and safe solution is to write only “Türkçe” in the list.

A third detail is what icons call to mind. The piggy bank used to convey a savings function is an image that can cause discomfort in the Turkish market, and it is generally replaced with a money box or a safe icon. Hand gestures likewise call for care; some gestures carry different meanings from one culture to another.

What this means for you

If you are having a Turkish version of a product prepared, do not leave the icons outside the scope of the translation. Even where the translator carries the text across correctly, reporting the disconnect between label and icon is possible only if it is asked of them. That request can be written into the brief in a single sentence and brings you no extra cost at all. What it does bring is shorter rounds of correction.

1. Share screenshots along with the translation file; translation without context cannot see an icon problem.
2. Ask the translator to report icon-label mismatches in a separate list.
3. If you use flags in the language selector, move to language names.
4. Review icons with risky cultural associations before release.
5. Keep a text label for every icon; both localisation and accessibility gain by it.

For a scope in which interface texts and visual elements are handled together you can look through our services page, and for the files you have to hand you can request a quote.

Frequently asked questions

What does a broken metaphor mean?

It is the disappearance in the target language of the analogy between an icon and the word that names it. The disconnect between the house icon and the phrase “ana sayfa” is the typical example.

Should icons be localised too?

Where necessary, yes. The visual-iconic level is one of the dimensions subject to cultural adaptation. Icons that depict objects the target user does not recognise can be changed.

Is it inadvisable to use flags in language selection?

Generally it is. A flag represents a country, not a language. Where the same language is spoken in more than one country, a portion of the users is left out.

How is the decision to change an icon reached?

First the icon is tested with a small group of users to see whether it is understood. Icons with high recognisability are kept; where the link of meaning has snapped, a label is added or the icon is changed.

References

• Jiménez-Crespo, M. A. (2024). Localization in Translation. Routledge, pp. 101-109.
• Yunker, J. (2010). The Art of the Global Gateway. Byte Level Books.
• Dietz, F. (2006). Issues in localizing computer games. Perspectives on Localization, John Benjamins.