Five More Privileges of English Speakers, part 2: Language and Software

For the previous part in the series, see Five Privileges of English Speakers, part 1.

I’m continuing the series of posts in each of which I write about five privileges that English speakers have without giving it a lot of thought. The examples I give mostly come from my experience translating software, Wikipedia articles, blog posts, and some other texts between English, Hebrew, and Russian. Hebrew and Russian are the languages I know best. If you have interesting examples from other languages, I am very interested in hearing them and writing about them.

I’m writing them mostly as they come into my mind, without a particular order, but the five items in this part of the series will focus on usage of the English language in software, and try to show that the dominance of English is not only a consequence of economics and history, but that it’s further reinforced by features of the language itself.

1. Software usually begins its life in English

English is the main language of software development worldwide.

The world’s best-known place for software development is Silicon Valley, an English-speaking place. That’s the place of Facebook, Google, Apple, Oracle and many others. California is also the home of Adobe.

There are several other hubs of software development in United States: Seattle (Microsoft, Amazon), North Carolina (Red Hat), New York (IBM, CA), Massachusets (TripAdvisor, Lotus, RSA), and more. The U.S. is also the source for much of computer science research and education, coming from Berkeley, MIT, and plenty of other schools. The U.S. is also the birthplace of the Internet, originally supported by the U.S. Department of Defense and several American universities. The world wide web, which brought the Internet to the masses, was created in Switzerland by an English speaker.

Software is developed in other countries—India, Russia, Israel, France, Germany, Estonia, and many other countries. But the dominance of the U.S. and of the English language is clear. The reason for this is not only that the U.S. is the source for much of computer technologies, but also—and probably more importantly—that the U.S. is the biggest consumer market for software. So developers in all countries tend to optimize the product for the highest-paying consumers, and these only need English.

When engineers write the user interface of their software in English, they often do not give any thought to other languages at all, or make translation possible, but complicated by English-centric assumptions about number, gender, text direction, text size, personal names, and plenty of other things, which will be explored in further points.

2. Terminology

English is also the source for much of the computer world’s terminology. Other languages have to adapt terms like smartphone, network, token, download, authentication, and thousands of others.

Some language communities work hard to translate them all meticulously into native words; Icelandic, Lithuanian, French, Chinese, and Croatian are famous examples. This is nice, but requires effort on behalf of terminology committees, who need to keep up with the fast pace of technological development, and on behalf of the software translators, who have to keep with the committees.

Some just transliterate most of them: keep the term essentially in English, but rewritten in the native alphabet. Hindi and Japanese are examples of that. This seems easy, but it is based on a problematic assumption: that the target language speakers who will use the software know at least some English! This assumption is correct for the translators, who don’t just know the English terms, but are probably also quite accustomed to it, but it’s not necessarily correct for the end users. Thus, the privilege is perpetuated.

Some languages, such as Hebrew, German, and Russian, are mid-way, with language academics and purists pulling to purer native language, engineers pulling to more English-based words, and the general public settling somewhere in between—accepting the neologisms for some terms, and going for English-based words for others.

For the non-English languages it provides fertile ground for arguments between purists and realists, in which the needs of the actual users are frequently forgotten. All the while, English speakers are not even aware of all this.

3. Easy binary logic word formation

One particular area of computer terminology is binary logic. This sounds complicated, but it’s actually simple: in electronics and software opposite notions such as true / false, success / failure, OK / Cancel, and so forth, are very common.

This translates to a great need for words that express opposites: enable / disable, do / undo, log in / log out, delete / undelete, block / unblock, select / deselect, online / offline, connect / disconnect, read / unread, configured / misconfigured.

Notice something? All of the above words are formed with the same root, with the addition of a prefix (un-, dis-, de-, mis-, a-), or with the words “on” and “off”.

A distinct, but closely related need, is words for repetition. Computers are famously good at doing things again and again, and that’s where the prefix re- is handy: reconnect, retry, redo, retransmit.

These features happen to be conveniently built into the English language. While English has extremely simple morphology for declension and conjugation (see the section “Spell-checking” in part 1 of the series), it has a slightly more complex morphology for word formation, but it’s still fairly easy.

It is also productive. That is, a software developer can create new words using it. For example, the MediaWiki software has the concept of “oversight”—hiding a problematic page in such a way that only users with a particular permission can read it. What happens if a page was hidden by mistake? Correct: “unoversight”. This word doesn’t quite exist elsewhere, but it doesn’t sound incorrect, because familiar English word formation rules were used to coin it.

As it always happens, English-speaking software engineers either don’t think about it at all, or think that other languages also have similar word formation rules. If you haven’t guessed it already, it is not true. Sime other European languages have similar constructs, but not necessarily as consistent as in English. And for Semitic languages like Hebrew it’s a disaster, because in Semitic languages prefixes are used for entirely different things, and the grammar doesn’t have constructs for repetition and negation. So when translating software user interface strings into Hebrew, we have to use different words as opposites. For example the English pair connect / disconnect is translated as lehitḥabér / lehitnaték—completely different roots, which Hebrew is just lucky to have. Another option is to use negative words like lo and bilti, or bitul, but they are often unnatural or outright wrong. Having to deal with something like “Mark as unread” is every Hebrew software translator’s nightmare, even though it sounds pretty straightforward in English.

English itself also has pairs of negative words that are not formed using the above prefixes, for example next / previous and open / close, but in many other languages they are much more common.

4. Verbing

“Verbing weirds language”, as one of the famous Calvin and Hobbes panels says.

Despite being a funny joke in the comic, it’s a real feature of the English language: because of how English morphology and syntax work, nouns can easily jump into the roles of adjectives and verbs without changing the way they are written.

For English, this is a useful simplification, and it works in labeling, as well as in advertising. “Enjoy Coca-Cola” is something more than an imperative. The fact that it’s a short single word and that it’s the same in all genders and numbers, makes it more usable as a call to action than it would be in other languages. And, other than advertising, where are calls to action very common? Software, of course. When you’re trying to tell a user to do something, a word that happens to be both the abstract concept and the imperative is quite useful.

Perhaps the most famous example of this these days is Facebook’s “Like”. Grammatically, what is it in English? Imperative? A noun describing an abstract action? Maybe a plain old noun, as in “chasing likes” (this is a plural noun—English verb don’t have a plural form!)? Answer: it’s all of them and more.

When translated to Hebrew in Facebook’s interface, it’s Ahávti, which literally means “I loved it”. Actually, this translation is mostly good, because it’s understandable, idiomatic, and colloquial enough without compromising correctness. Still, it’s a verb, which is not imperative, and it’s definitely not a noun, so you cannot use it in a sentence as if it was a noun. Indeed, Hebrew speakers are comfortable using this button, but when they speak and write about this feature, they just use its English name: “like” (in plural láykim). It even became a slightly awkward, but commonly used verb: lelaykék. Something similar happens in Russian.

It would be impossible in Hebrew and Russian to use the exact same word for the noun and the verb, especially in different persons and genders. Sometimes the languages are lucky enough to be able to adapt an English verb in a way that is more or less natural, but sometimes it’s weird, and hurts the user experience.

5. Word length

This one is relatively simple and not unique to English, but should be mentioned anyway: English words are neither very long, nor very short. Examples of languages where words are, on average, longer than in English, are Finnish, Tamil, German, and occasionally Russian. Hebrew tends to be shorter, although sometimes a single English word has to be translated with several Hebrew words, so it can get also get longer. This is true for a pretty much any language, really.

In designing interfaces, especially for smaller screens, the length of the text is often important. If a button label is too long, it may overflow from the button, or be truncated, making the display ugly, or unusable, or both.

If you’re an English speaker, it probably won’t happen with you, because almost all software is usually designed with the word length of your language in mind. Other languages are almost always an afterthought.

The good practice for software engineers and designers is to make sure that translated strings can be longer. Their being shorter is rarely a problem, although sometimes a string is so short that the button may become to small to click or tap conveniently.

---

Generally, what can you do about these privileges?

Whoever you are, remember it. If you know English, you are privileged: Software is designed more for you than for people who speak other languages.

If you are a software engineer or a designer, at the very least, make your software translatable. Try to stick to good internationalization practices and to standards like Unicode and CLDR. Write explanations for every translatable string in as much detail as possible. Listen to users’ and translators’ complaints patiently—they are not whining, they are trying to improve your software! The more internationalizable it is, the more robust it is for you as a developer, and for your English-speaking users, too, because better design thinking will be going into each of its components, and less problematic assumptions will be made.