Of Bicycles, Biographies and Bits

My New Year's Resolution is off to shaky start.  I usually don't bother with such nonsense.  I mean, what should I resolve to do?  I resolve to continue to be perfect in all ways?  I was going to do that anyway and I was even planning to be humble.  This year is different.  I have been neglecting this fine blog due to school overload.  This year I resolve to write in my blog more and to ensure the posts have meaningful content.  So, hear goes.  This is a reprint of quick reflection paper from a class.  When I make reference to Walter Ong, I'm citing his 1982 book Orality and Literacy.  When I refer to Verran, I'm referring to Australian anthropologist Helen Verran who studies how human cultures store and recall knowledge.  The paper referred to describes an attempt at collaboration and Australian aborigines and ecologists in conducting prescribed burns.  I'll draw up the full citations later on.  The paper is a great read even for non academic readers.  I'm also drawing a lot of ideas from Karl Popper and Michel Foucault as well.
 

I work at an institution that develops language technology where I'm sort of a librarian, editor, tech writer, software developer, QC engineer, system administrator, amateur linguist and part time lexicographer. Now you can see why I might have trouble writing a cover letter for job hunting. This is part of the reason that even more graduate school seemed to appealing. This, my friends, is what happens when you never specialize in anything: You get jobs that take three pages to describe. These days, I work part time as I help my replacement's replacement's replacement settle in and deal with the backlog.

At any rate, one of our main products is conversational telephone speech. We collect it through a variety of perfectly legit means. We then transcribe, annotate, translate, treebank, spindle, rip tear and immolate that speech. From there, it gets used to develop, train, test and evaluate software systems that deal with language. Since I have to verify that everything meets some sort of quality standard before shoving it out the door, I wound up listening to a lot of the collected speech back when I worked full time. I would listen to make sure that the recordings were intelligible, contained the language advertised and that the speakers match up to the provided metadata. I'm not really fluent in any other language besides North American Geek, but since I grew up with and work around so many languages, I score better at language recognition tasks than most software systems do. I guess this is a sort of idiot savant ability that I can amaze friends with at parties. “Hmm... I detect the spicy tang of Highland Malay in your accent my friend.” or “Ha! It's a trick question. There is no standard form of spoken Klingon, but the digloss in Literary Klingon puts an entirely different meaning on that word”.

Couple of years back, I built up a sample corpus with which to impress conference goers and the like. Completing this corpus meant reviewing quite a bit of telephone speech. I have joked in the past that I work on a public sector, open source Echelon and there is some truth to that. We run a massive signals intercept operation that hoovers up news and other broadcast content to develop language corpora. One friend jokingly refers to me as “Condor” in reference to Robert Redford's famous part. Broadcast news is good for a lot of things, but to truly capture the human quality to speech, you need conversations, hence our efforts to collect telephone speech. One of our better ploys to obtain colloquial foreign language speech is to pay immigrants to call their friends and family back home. We capture the calls using what amounts to a fully disclosed wiretap. Once we capture the speech, we screen out any personally identifiable information and keep names to first names as much as humanly possible. We like to fool ourselves that we screen out the human element, but that's not the case. The machines don't care about this stuff when they try to parse our Natural Languages. The machines don't reflect in anything other than the electromagnetic spectrum.

The human QCing the recordings is a very different matter. Listening to the recordings, I was all too aware that I was eavesdropping on a private conversation. I have another ability beyond my idiot savant language recognition talent. This ability has far more practical applications. I can pick out emotional states in tone of voice in detail. Likewise, empty spaces in conversation, pregnant or barren, yield vast riches of information to me. There's so much that's not at all verbal contained in speech. An awareness of these modes of communication can rise to the level of functional telepathy. Once upon a time, this talent was focused on ascertaining the level of aggression in a speaker or the threat contained in any conversational occasion. Fortunately, I have broadened the talent in more recent years. This means that when I listen to people speaking in a foreign language, I can never divorce my emotional self from the part that evaluates recording quality and identifies the language spoken. The human element in all these conversations comes through loud and clear and I come away feeling as though these people are a set of invisible co-workers.

They're unseen, phantoms devoid of form, but all all too tangible. Listening to people from seven different countries call home on my workplace's dime, I came to the conclusion that foreigness is a cultural construct and that belonging is our natural state. Listening to Korean and Japanese people call, I would hear the happy reunion of good friends catching up with one and other's lives. And that should be no surprise. The reasons for these casually pleasant calls are economic, not national. Koreans and Japanese people in this country are elective emigrants or visitors. They came here hoping for something and were confident in their ability to purchase it. The story changes considerably with Arabic and Spanish speakers. For these two groups, there is a note of desperation and and weary relief to the reunions. The first “Allo! Allo!” might sound casual, the greeting that anyone in the Levant or the Gulf States might use on the telephone, but once the proper nouns start, the emotion pours out of the headphones. In other Arab countries, you might here “Allah akbar!” or even “Allo akbar!”, but the rest follows the pattern. These conversations are intimate. Yes, the callers know we're recording them, but, on their budgets, they have no other way to place a call. Maybe, the money saved on that call lets them eat an extra meal this week. The reasons for the pathos once again are not national, not cultural. Rather they are political and economic. A good many of the Arab speakers on my side of the Oceanic divide are involuntary emigrants. Their choice to leave home was forced on them by the circumstances. Some of those are financial. Many come here to escape a very real threat of violence either from the state or from rival tribes back home. One way or another, I had a tap into their emotional lifeline. One phone call to a Gulf State seemed to say all that mattered in a language anyone could understand. The conversation started off with the usual “Allo! Allo!” and then a name popped out accompanied with another word in Gulf Arabic. “Hi. Mom, Dad it's me...” the young woman said plainly to my ear. The proper nouns sighed forth in happy relief. People piled onto the call, some on a second line, others crowded around the first receiver. Mom, Dad, Granma, Auntie and Little Brother were all there. The conversation included a great many “I miss you, so”s, several “are you eating well?”s and consensus of “we're so happy for you”s. Like I said, I don't speak any of the Gulf dialects, I only know the language of the pauses, the dialect of the empty space and the register of the tempo. I know the words the ocean breeze uses when it blows through your far away home and says welcome back if only for a few minutes.