Linguistics for Software Engineers: Why Concatenation is a Bad Idea

noraneko — Sat, 19 Nov 2005 07:25:00 GMT

It is very tempting to write a application that concatenates strings or audio to make a single phrase or sentence. It seems like it would be really helpful if you are squeezed for space or to speed things up. However, natural language is a lot more complicated than you might think. Let's take a straightforward sentence.

The hero kills the monster with his iron sword.

If we change the sex of the character then we also have to change the noun and pronoun:

The heroine kills the monster with her iron sword.

If we change the number, we have to change the noun, verb, pronoun and number:

The heroes kill the monsters with their iron swords.

And that's just English.

Now, if we translate into French, we will also have to inflect the definite article to agree with the gender of the noun and the genitive construction is different from English too.

Les héros tuent les monstres avec leurs épées de fer.

And just to complicate this simple sentence, Swedish appends the definite article to the noun.

a hero en hjälte a heroine en hjältinna
the hero hjälten the heroine hjältinnan

Fun, isn't it? These are very limited examples, but I hope you get the picture. Natural languages are pretty darn complicated. If you plan to localize your software for other languages, just don't concatenate!

Translation versus Transliteration

noraneko — Tue, 15 Nov 2005 18:48:00 GMT

The issue of translating "proper nouns" comes up a lot in coding software that will be localized into other languages. Developers, in my experience, like to hardcode anything and everything if possible. This is not just a US programming phenomenon – I've seen Japanese developers do the same thing, making localizing the software into English just as difficult.

The argument is usually something like "You don't translate a company name, like Boeing." Ah, but you do transliterate it. In fact, Boeing, and Microsoft for that matter, transliterate their company names on their own websites. When transliterated phonetically into Japanese katakana Microsoft becomes "マイクロソフト" and Boeing becomes "ボーイング " Conversely, since most non-Japanese don't read Japanese, we are used to reading Japanese transliterated into the Roman alphabet, e.g. "Tokyo" than "東京". There are plenty of languages which do not use the Western European writing system.

Just to be clear, here are the definitions of translate and transliterate:
Translate: turn words into different language: to reproduce a written or spoken text in a different language while retaining the original meaning.
Transliterate: transcribe something into another alphabet: to represent letters or words written in one alphabet using the corresponding letters of another.
Microsoft® Encarta® 2006. © 1993-2005 Microsoft Corporation. All rights reserved.

What this practically means that any text that is displayed to a user must be able to be localized, whether it's translated or transliterated. Plain and simple.

Noraneko's Multilingual Musings : Int'l Software Development

Linguistics for Software Engineers: Why Concatenation is a Bad Idea

Translation versus Transliteration