Monthly Archives: June 2014

Translation tech: We’re getting there…

Language barriers seem to be frustrating us more than ever before. What cognitive dissonance it is that in today’s globalized world we can get in touch so easily but still be unable to communicate. For this reason Rosetta Stone is a $300-million company and growing every year, and language-instruction apps like Duolingo and MindSnacks are coming out all the time.

We have fantasies of Matrix-like brain implants that give you instant fluency in any language you’d like to install. Nearly as futuristic but actually existent is Word Lens, an app that translates written language that you point to with your smartphone’s camera (and actually works quite well). The company was recently bought by Google, who is implementing the technology in their Glass productOur dreams of instant communication with speakers of other language are still far off, but we’ve certainly come a long way from pocket phrasebooks.

MuuziiNext up is Muuzii, which promises to be an instant, SMS-based translation service. The basic use-case is this: Two people who don’t speak the same language both have basic cellphones. Person One types a message in his language to Muuzii, and Muuzii quickly responds with the message translated to Person Two’s language. Person One can then forward or sound out the message to get it to Person Two. It’s a subscription-based service (quite reasonable at less than $5/month) that currently works on AT&T but is planned for all U.S. carriers soon.

They have a promotion going on right now where you can try out the service for free from any carrier. I tried it out, and I was somewhat disappointed. Here’s why:

As I see it, Muuzii has three main claims to fame. It follows through on two of them, but flops on the third. One: It’s meant to be fast, and it certainly is. In my tests, the response always came within a second or two. Two: Dumbphones welcome. Since it’s an SMS-based service, you don’t need a smartphone. That’s key; we like to think that everyone has a smartphone these days, but that’s just not the case.

Three: There’s a real, live person doing the translating (supposedly). Here’s where Muuzii doesn’t live up to its potential—at least not yet. The problem with machine-based translators is that they’re not infallible. They generally aren’t good with casual sentences, slang and idioms, first of all. And what if you spell something wrong? What if you tried to write something, but inadvertently spelled a different word? An actual person would see what you meant, but not a machine. I was really excited to hear that Muuzii has “a robust team of skilled linguists on-call 24/7,” because such a team could translate so much better than a machine. But in practice, Muuzii can’t seem to translate spelling errors. In my test, I sent the query:

did the untied sttates win the game agains Germanny?

And Muuzii replied in Spanish:

¿los sttates desatados ganó el juego agains Germanny?

The translation is interesting. Muuzii caught that sttates was meant to be plural, even if it couldn’t manage to translate it. Strangly, it gave me the verb conjugation ganó, which is for third-person singular, when it had already identified the subject as plural. It correctly translated untied (plural) as desatados, but it didn’t realize that I actually meant united rather than untied.

I believe a “robust team of skilled linguists” would have been able to translate my original sentence perfectly. Maybe they would have included fun spelling errors of their own—but that’s a discussion for a class on translation theory. Perhaps it’s the case that Muuzii isn’t showcasing the true extent of their product during this free trial (which would be a shame, because then what’s the point?), or maybe they’re simply over-promising.

When acronyms should not be capitalized

Acronyms, abbreviations in which each letter is the first letter of a word in a phrase, are customarily capitalized. But not always. Laser, for example, is an acronym that stands for light amplification by stimulated emission of radiation; the word is now so common that nobody thinks of it as an acronym, and subsequently nobody capitalizes it. This seems to be governed by the same mechanism as the disappearance of hyphens in compound words: as words get more common, they tend toward the style of normal running text—that is, lowercase.

Benelux is another acronym, standing for Belgium Netherlands Luxembourg, but it’s not written BeNeLux. A growing convention, which I first came across in Robert Bringhurst’s The Elements of Typographic Style, dictates that if an acronym is pronounced as a word, then it should not be written in all caps. (As opposed to initialisms, which are pronounced as a series of letters. Compare Nato to USA.)

Still, plenty of people capitalize NATO and AIDS and NASA, and I don’t think anyone would argue that doing so is incorrect. (People have certainly argued about stranger things, though.) But one case comes to mind wherein capitalizing certain acronyms is seen as gauche at best: The neographic texting/Internet shortcuts, most notable among them lol.

22344_250266667263_5206655_nPeople who capitalize LOL have long been the subject of Internet ridicule. There is no shortage of forum messages, Facebook pages, tweets, or blog posts about the practice. The Internet community, so varied in other respects, seems to be in agreement here. And just as Apple iOS’s autocorrect feature has caused strife for other reasons, people seem to hate that it capitalizes LOL for them. Why is this?

The consensus is that capitalizing LOL is the purview of old people and other e-newbies. And no millennial worth his weight in iPhones wants to be seen in this light. “We get why you people might think it should be capitalized like a normal acronym,” the collective voice of the Internet seems to say, “but you need to get with the program and realize it’s not.” It’s quite interesting that the convention evolved such that this acronym (along with others, like ikr, idk, btw) should be treated quite differently than normal.

What’s going on here? The preference stems from the pervasive practice of writing casual text in all lowercase. Casual text values simplicity and speed, and pressing that shift button is seen as unnecessary. In such environments, sentence starters, proper nouns—even the letter I—tend to be written in lowercase. Given that such acronyms are found almost exclusively in casual contexts, it makes sense that it would need to be written in lowercase as well.

It seems lol is here to stay, and it’s being used in more and more places. It’s crawled from the trenches of Internet subculture to some less-important business emails, which has been quite a journey. I don’t foresee that it’ll go much further, but who knows. In any case, it does seem to be shifting a bit as it goes. For example, I am seeing Lol more and more often: The first L is capitalized when the word begins a sentence; in other words, we see this neographic token in environments where not all orthographic norms are flouted (we relent to capitalize the first letter in each sentence, but no way are we going to capitalize LOL).

On a somewhat unrelated note, you might be interested in the equivalents of lol in other languages.

As spell-check evolves, what are we losing?

Spell-check has been around for decades, and despite some improvements, it still leaves a lot to be desired. Simply put, it doesn’t always know what’s best. On the iPhone, for example, the autocorrect feature is simply blunderful. And it still can’t really help us when we type to instead of too

Ginger is bringing spell-check to the next level. Not only does it offer spelling and grammar checking, but its newest software offers a Sentence Rephraser. (Incidentally, rephraser is marked as a spelling error in Ginger. They must have meant to name it Sentence Refresher.) It’s a pretty sophisticated language processor that relies upon a corpus of “high-quality” Web writing to offer context-sensitive suggestions. As an aside, I think this software wins where iOS’s autocorrect fails because it makes suggestions rather than auto-corrections.

As quoted in a recent VentureBeat article, Ginger CEO Maoz Schact says, “We see ourselves as trying to raise the level of English from the pre-mobile days. We’re able to make you look good on your mobile with minimal effort” (emphasis mine).

Of course, when he says “pre-mobile” days, he really means “pre-mobile-post-print” days. As I’ve written about before, the obsession with spelling and grammatical correctness only came about in the past few hundred years. Before the printing press was invented, we were a lot more lenient.


This is an interesting, if predictable, reaction to the text-speak, typos and other casual constructions that have arisen in the speed-over-accuracy mobile age. True, it’s a natural extension of the iPhone (and maybe Android… I’m not familiar enough) autocorrecting r to are and u to you. It shows that, at least among certain groups, such neographic shortcuts are still looked upon disparagingly.

At least from some perspectives, and in some places. I, for one, find it overly pedantic when someone on IM insists on capitalizing every proper noun and including all their periods, and I know a lot of other people do, too. Such “errors” are not simply errors—to regard them as such would be to take too limited a view. They are actually information in themselves: They indicate the register (that is, the level of formality) of the speech, reflecting the relationship between the interlocutors, and they offer important contextual information. For example, if you receive an IM from your boss that’s rife with spelling errors and has no punctuation, you probably wouldn’t conclude that he’s a dumbo who doesn’t know how to spell follow-up, but rather that he’s in a big hurry and whatever he’s asked you for is urgent.

If we try to be “correct” at all times, we lose out on all this paralinguistic information. And that’s a shame: In speech, we can rely on speed, tone, facial expressions and gestures to provide this type of information, but in writing we have a lot less to go on. When we don’t have access to bold and italics, we (perhaps unconsciously) rely on “errors” to convey deeper shades of meaning. So in some cases, being “correct” can actually lead to misunderstanding. What we have to remember is that the type of writing that’s best always depends on the context. It doesn’t matter who you are: Diction appropriate for an academic paper is not appropriate in a text message to your significant other. In fact, language that feels overly formal for a given situation is likely to be interpreted as distancing. For example, we understand that sentences are “supposed” to end in periods, but if you end your text messages in periods, you’ll probably come off as angry.

Ginger wants to move from a standalone app to an OS-pervasive helper (and in Android, it already is), but that may not be for the best. Ultimately, it’s up to us humans to modulate our language… and, as in so many other instances, technology that we attempt to use as a shortcut might only screw things up.