5.15.2005

eh, so so expectations.

I got a text message that said "i hope garden state isn't cunning you out," because "c" shares a phone button with "b" and "n" shares a button with "m." And I immediately knew that the person sending the message was using one of those T-9 Word things where you press each button once, and the phone displays the most likely word formed from each of those buttons. "if" also often reads "he" in this system.

Additionally, I was watching CNN with closed-captioning, and in a discussion that was obviously about "rumors," the closed-captions displayed the word "roomers." I don't know exactly how those systems for generating real-time captions work, but from this one example, we can assume that it's based in phonetics, like a stenographer's machine that's computer-translated into English orthography.

One more example: we often read "/" at the end of a sentence as "?" because the two share the same key, and to switch from the key's shifted option to the non-shifted option is a common typo on a standard qwerty keyboard.

Here's the basic pattern of these observations: the writer's intended meaning is known; the writer deviates from that meaning; the form of the error reveals the system of inputs the writer is using to create the message. The field of linguistics often uses this method. When a native Spanish speaker says "eschool" when we understand that he means to say "school," we can begin to infer patterns in his system of inputs, which should correspond to Spanish phonetics. And there's the problem with linguistics as an empirical study: the speaker (or writer)'s intended meaning has to be known when an observation is made.

This method seems to function fine for phonetics, but when we reach semantics, we can no longer use errors in meaning to predict the speaker's system for creating meaning, because our method is founded on the idea that we know what the speaker means, even if he doesn't say it. For example: I stayed with a German girl who kept telling me "you must not do this, you must not do that," and I thought she was being awfully harsh with me about these minor things. But come to find out, she meant "you don't HAVE to do this or that," because that's the way those modals translate into German. She was defying my expectations in a systematic way, but I had no way of knowing, because I didn't know her intended meaning separate from what she actually said.

There's always expectation involved in these types of studies. There's no way to begin at the beginning with language.

2 comments:

libcat said...

I don't know exactly how those systems for generating real-time captions work, but from this one example, we can assume that it's based in phonetics, like a stenographer's machine that's computer-translated into English orthography.

Actually, they're human transcribed, rather like stenography itself. See, for example, the trouble CBS got into a few years ago when Dave Letterman said "damn" or some four-letter word on his show---it was bleeped for the hearing audience, but initial broadcasts had the word printed in the captioncast. Joe Clark talks about one instance here.

Blogger's comment system refuses both <blockquote> and <q>. I am disappointed.

Cristi said...

Thanks for the clarification. I just assumed a human would notice something odd like "roomers" but people are busy...

Also, I'm disappointed in blogger's comment system too. We obviously use the blockquote tag a lot in the posts themselves.

Related Posts with Thumbnails