There is a common critique of journalists that they often let an internal narrative color their reporting, to the point where they simply parrot back the narrative in their head rather than report the facts on the ground (see here for a discussion of this).
My hometown of Chico got its spotlight in the sun recently because its favorite son Aaron Rogers is the star quarterback of the Packers about to play in the Super Bowl. Unfortunately, the NYT's article is a near perfect example of journalists letting a narrative do the talking when the facts blatantly contradict their claims:
The usual slang words like awesome or cool are not heard much. Nice is in. As in: “You won the lottery? Nice.”
The narrative this spins is that small towns are all Mayberrys where everyone is pure and innocent and righteous and better than them damned city-folk. It has been evoked routinely in political reporting.
I'm a Chico boy. I graduated from Chico jr. High and walked across the street and graduated from Chico High then walked across the street and graduated from Chico State*. And I can assure you that awesome and cool are every bit as frequent there as anywhere else (personally, I had an unhealthy fondness for hella back in 1987). And believe me, if you won the lottery in Chico, no one would say nice. They would say, "No fukkin way! No fukkin way! Really! No fukkin way!" ... just like everywhere else.
*no joke, those three schools are literally across the street from each other.
Monday, January 31, 2011
how (not) to do linguistics
Jonah Lehrer, the neuro-blogger, has a mixed track record, as far as I'm concerned. His initial blogging was nice, but a tad lightweight, then he started to sound a bit too Malcom Gladwell-ee (in that I wasn't entirely sure he knew what he was talking about beyond having a few short phone calls with one or two scientists then babbling on about a topic).
But he's hit a home run with this long New Yorker piece about the failure of the journal review process in science: The Truth Wears Off. He draws examples from medicine, physics, and psychology.
Perhaps the most disappointing part is the realization the the standards of testing and conclusiveness in linguistics are so far from those in more established science.
Before the effectiveness of a drug can be confirmed, it must be tested and tested again. Different scientists in different labs need to repeat the protocols and publish their results. The test of replicability, as it’s known, is the foundation of modern research. Replicability is how the community enforces itself. It’s a safeguard for the creep of subjectivity.
Repeating studies is virtually unheard of in linguistics. Also, Lehrer mentions the publication bias in journals. When a result is discovered, there is a bias towards positive results. After a while, once the result is accepted, then only negative results are published because only that is "interesting" anymore. But I would expand this point to say this same bias exists at every stage of the research process. We want to find things that happen, we don't care about spending 5 years and thousands of hours discovering that X does NOT cause Y! So when young grad students begin scoping out a new study, they throw away anything that doesn't seem fruitful, where fruitful is defined as yielding positive results. This bias affects the very foundation of the research process, namely answering the basic question: what should I study?
As a side note, engineers seem perfectly happy to follow through on null results. They need to know the full scope of their problem before solving it. Scientists can learn a lot from engineers (and vice versa).
[Psychology professor Jonathan] Schooler recommends the establishment of an open-source database, in which researchers are required to outline their planned investigations and document all their results. “I think this would provide a huge increase in access to scientific work and give us a much better way to judge the quality of an experiment,” Schooler says. “It would help us finally deal with all these issues that the decline effect is exposing.”
Coincidentally, I was recently tweeting with moximer and jasonpriem about this and we agreed that research wikis are worth explolring. My vision would be something akin to Wikipedia but where a researcher stores all of their data, stimuli, results, etc, finished or not. The data could be tagged as tentative, draft, failed, successful, etc. As the research goes on, the data get updated. Not only would this record failure (which, as Leherer points out in the article) is as valuable as success, it also records change. How did a study evolve over time?
True, the data would become huge over time across many disciplines, but that just means means we need better and better data mining tools (and the boys at LingPipe are working away at those tools).
HT rapella
But he's hit a home run with this long New Yorker piece about the failure of the journal review process in science: The Truth Wears Off. He draws examples from medicine, physics, and psychology.
Perhaps the most disappointing part is the realization the the standards of testing and conclusiveness in linguistics are so far from those in more established science.
Before the effectiveness of a drug can be confirmed, it must be tested and tested again. Different scientists in different labs need to repeat the protocols and publish their results. The test of replicability, as it’s known, is the foundation of modern research. Replicability is how the community enforces itself. It’s a safeguard for the creep of subjectivity.
Repeating studies is virtually unheard of in linguistics. Also, Lehrer mentions the publication bias in journals. When a result is discovered, there is a bias towards positive results. After a while, once the result is accepted, then only negative results are published because only that is "interesting" anymore. But I would expand this point to say this same bias exists at every stage of the research process. We want to find things that happen, we don't care about spending 5 years and thousands of hours discovering that X does NOT cause Y! So when young grad students begin scoping out a new study, they throw away anything that doesn't seem fruitful, where fruitful is defined as yielding positive results. This bias affects the very foundation of the research process, namely answering the basic question: what should I study?
As a side note, engineers seem perfectly happy to follow through on null results. They need to know the full scope of their problem before solving it. Scientists can learn a lot from engineers (and vice versa).
[Psychology professor Jonathan] Schooler recommends the establishment of an open-source database, in which researchers are required to outline their planned investigations and document all their results. “I think this would provide a huge increase in access to scientific work and give us a much better way to judge the quality of an experiment,” Schooler says. “It would help us finally deal with all these issues that the decline effect is exposing.”
Coincidentally, I was recently tweeting with moximer and jasonpriem about this and we agreed that research wikis are worth explolring. My vision would be something akin to Wikipedia but where a researcher stores all of their data, stimuli, results, etc, finished or not. The data could be tagged as tentative, draft, failed, successful, etc. As the research goes on, the data get updated. Not only would this record failure (which, as Leherer points out in the article) is as valuable as success, it also records change. How did a study evolve over time?
True, the data would become huge over time across many disciplines, but that just means means we need better and better data mining tools (and the boys at LingPipe are working away at those tools).
HT rapella
Friday, January 28, 2011
the sociolinguistics of height in China
Ingrid at Language on the Move has some thoughtful comments on the relationship between height and learning English in China. If you're under 1.6 meters, forget it. There are subtle but very real socioeconomic barriers in your way. Money quote:
I have supervised research related to English language learning and teaching in China for almost a decade and have read most of the research on the topic published in English. However, never before have I come across the importance of height. I take this as evidence for the importance of doing ethnographic research. Otherwise, what is the point of doing sociolinguistic research if you can’t discover anything you hadn’t already decided in advance would be important?!
I taught EFL in China back in 1998 at a private school in Guangzhou catering mostly to working professionals. Much has changed since then, as China has changed so much. I don't recall ever talking about height as a factor, but certainly cost and hours were a significant issue that made it virtually impossible for any poor workers to consider taking English classes. As a 6 foot 4 blond American, though, I was treated like a rock star. It was kinda weird.
I have supervised research related to English language learning and teaching in China for almost a decade and have read most of the research on the topic published in English. However, never before have I come across the importance of height. I take this as evidence for the importance of doing ethnographic research. Otherwise, what is the point of doing sociolinguistic research if you can’t discover anything you hadn’t already decided in advance would be important?!
I taught EFL in China back in 1998 at a private school in Guangzhou catering mostly to working professionals. Much has changed since then, as China has changed so much. I don't recall ever talking about height as a factor, but certainly cost and hours were a significant issue that made it virtually impossible for any poor workers to consider taking English classes. As a 6 foot 4 blond American, though, I was treated like a rock star. It was kinda weird.
chomsky and performance art
Artist Annie Dorsen has created a chatbot performance piece around the debate between Noam Chomsky and Michele Foucault on Dutch TV in 1971 (videos here).
This snippet is mildly interesting, but I couldn't help wondering what technology was used, especially the speech synthesis, because, frankly, it's a bit clunky and old-fashioned. Perhaps that's part of the point. The computer screens appear to be running DOS shells too. Nothing wrong with that, purists will likely prefer it even, but combined with the clunky speech, the performance appears to be trading on a very outdated computational linguistic aesthetic. Truly the desert of the real? (okay, I had to through in some Baudrillard)
This snippet is mildly interesting, but I couldn't help wondering what technology was used, especially the speech synthesis, because, frankly, it's a bit clunky and old-fashioned. Perhaps that's part of the point. The computer screens appear to be running DOS shells too. Nothing wrong with that, purists will likely prefer it even, but combined with the clunky speech, the performance appears to be trading on a very outdated computational linguistic aesthetic. Truly the desert of the real? (okay, I had to through in some Baudrillard)
Thursday, January 27, 2011
can a machine learn jazz?
There's a contest dedicated to trying to answer that question: ISMIS 2011 Contest: Music Information Retrieval.
Computer scientists and engineers have long used contests and bake-offs to stimulate cutting edge research in linguistics (e.g., MUC), but linguists have lagged in this department. You rarely if ever hear about contests that pit one linguistic theory against another using a standardized data set (or maybe I've just missed them).
Nobel prize winning economist Joseph Stiglitz argues here that prizes are good for stimulating academic research. I agree whole heartedly and would like to see more direct competition between theorists. Exactly how a contest would be constructed is up for debate (I have a vague memory of some group trying to devise criteria by which to evaluate linguistic theories, maybe out of UCLA, but I can't seem to track it down; it's a remarkably difficult Google query to form).
HT: Jochen L. Leidner
Computer scientists and engineers have long used contests and bake-offs to stimulate cutting edge research in linguistics (e.g., MUC), but linguists have lagged in this department. You rarely if ever hear about contests that pit one linguistic theory against another using a standardized data set (or maybe I've just missed them).
Nobel prize winning economist Joseph Stiglitz argues here that prizes are good for stimulating academic research. I agree whole heartedly and would like to see more direct competition between theorists. Exactly how a contest would be constructed is up for debate (I have a vague memory of some group trying to devise criteria by which to evaluate linguistic theories, maybe out of UCLA, but I can't seem to track it down; it's a remarkably difficult Google query to form).
HT: Jochen L. Leidner
the linguistics of heaven and hell
The value of pop culture data for legitimate research is being put to the test. Exactly what, if anything, can the reality show Big Brother tell us about language change over time?
Voice Onset Time is a measure of how long you wait to begin vibrating your vocal folds after you release a stop consonant. Voiced stop consonants like /b/ and /d/ require two things: 1) stop all airflow from escaping the airway by closing the glottis and 2) after the air is released, begin vibrating the glottis (by using the rushing air). For non-linguists, think of a garden hose. Imagine you use your thumb to stop the water for a second and you let the pressure build, then you let go and water rushes out, but then you use your thumb to clamp down just a bit on the water to spray it. This is kinda like the speech production of voiced stop consonants in human language.
Though I’m no phoneticist, I really like VOT as a target of linguistic study for one crucial reason: it’s a clear example of a linguistic feature that varies according to your human language system but which you do NOT have conscious control over. What that means is that you cannot consciously change the length of your own personal VOT. Go ahead, try it. Make your VOT 20 milliseconds longer. Go ahead, I’ll wait…
Of course you can’t. Well, not consciously, but what researchers have found is that your brain, quite independent of conscious will or knowledge, can! Lab studies have found that people will unknowingly alter their VOTs according to certain situations, and the results are predictable. For example, they found that when listening to a set of long VOT stimuli, subjects will begin to lengthen their own VOTs, in essence accommodating the longer VOTs. Over the longer term it has also been shown that people will lengthen their VOT over their lifetime to accommodate cultural shifts. It has been shown that The Queen Mother herself had a longer VOT in her later life than during her younger days (few other people have been recorded consistently over a long period to provide such valuable data).
Here’s what Bane et al. did: They took recordings of confessional sequences from the UK reality TV show Big Brother (where groups of strangers are made to live with each other and occasionally speak to a camera alone like a video diary) and tested what happened to 4 crucial individuals (the ones that stayed on the show long enough to provide several months worth of data points). What they found was that their VOTs did in fact change, though no linear pattern was discovered (i.e., they did not simply get longer in a steady line). This paper is labeled as a progress report because they don't have a firm hypothesis about what actually is happening. Nice trick there boys, ;)
They did find one interesting thing: During part of the show, the house mates were physically divided into basically a caste system where half the people were low caste and half were high (a heaven and a hell). And this seemed to have an effect on VOT as well (sociolinguists are slap happy about this, I'm sure).
I haven’t looked at the actual numbers very closely, but in section 6, they say “Housemate trajectories seem to diverge when the divide is present…” However, just taking a glance at the Figure 3, it looks like they diverge at the beginning, then converge at the end, episode 65 (and remain somewhat similar until several episodes of non-DIVIDE have gone by). If my cursory glance is correct, I would assume it takes awhile for the convergence to manifest, and then it persists for awhile after DIVIDE is gone. But this is just me looking at the picture, not the actual data.
Finally, and this is just a readability point, but I would order the names in Figure 3 in the same order as the end point of each trajectory, making it easier to follow who is doing what.
Max Bane, Peter Graff, & Morgan Sonderegger (2011). Longitudinal phonetic variation in a closed system Linguistic Society of America 2011 Annual Meeting.
Voice Onset Time is a measure of how long you wait to begin vibrating your vocal folds after you release a stop consonant. Voiced stop consonants like /b/ and /d/ require two things: 1) stop all airflow from escaping the airway by closing the glottis and 2) after the air is released, begin vibrating the glottis (by using the rushing air). For non-linguists, think of a garden hose. Imagine you use your thumb to stop the water for a second and you let the pressure build, then you let go and water rushes out, but then you use your thumb to clamp down just a bit on the water to spray it. This is kinda like the speech production of voiced stop consonants in human language.
(image from Kval.com)
Though I’m no phoneticist, I really like VOT as a target of linguistic study for one crucial reason: it’s a clear example of a linguistic feature that varies according to your human language system but which you do NOT have conscious control over. What that means is that you cannot consciously change the length of your own personal VOT. Go ahead, try it. Make your VOT 20 milliseconds longer. Go ahead, I’ll wait…
Of course you can’t. Well, not consciously, but what researchers have found is that your brain, quite independent of conscious will or knowledge, can! Lab studies have found that people will unknowingly alter their VOTs according to certain situations, and the results are predictable. For example, they found that when listening to a set of long VOT stimuli, subjects will begin to lengthen their own VOTs, in essence accommodating the longer VOTs. Over the longer term it has also been shown that people will lengthen their VOT over their lifetime to accommodate cultural shifts. It has been shown that The Queen Mother herself had a longer VOT in her later life than during her younger days (few other people have been recorded consistently over a long period to provide such valuable data).
Here’s what Bane et al. did: They took recordings of confessional sequences from the UK reality TV show Big Brother (where groups of strangers are made to live with each other and occasionally speak to a camera alone like a video diary) and tested what happened to 4 crucial individuals (the ones that stayed on the show long enough to provide several months worth of data points). What they found was that their VOTs did in fact change, though no linear pattern was discovered (i.e., they did not simply get longer in a steady line). This paper is labeled as a progress report because they don't have a firm hypothesis about what actually is happening. Nice trick there boys, ;)
They did find one interesting thing: During part of the show, the house mates were physically divided into basically a caste system where half the people were low caste and half were high (a heaven and a hell). And this seemed to have an effect on VOT as well (sociolinguists are slap happy about this, I'm sure).
I haven’t looked at the actual numbers very closely, but in section 6, they say “Housemate trajectories seem to diverge when the divide is present…” However, just taking a glance at the Figure 3, it looks like they diverge at the beginning, then converge at the end, episode 65 (and remain somewhat similar until several episodes of non-DIVIDE have gone by). If my cursory glance is correct, I would assume it takes awhile for the convergence to manifest, and then it persists for awhile after DIVIDE is gone. But this is just me looking at the picture, not the actual data.
Finally, and this is just a readability point, but I would order the names in Figure 3 in the same order as the end point of each trajectory, making it easier to follow who is doing what.
Max Bane, Peter Graff, & Morgan Sonderegger (2011). Longitudinal phonetic variation in a closed system Linguistic Society of America 2011 Annual Meeting.
Wednesday, January 26, 2011
more jobs for linguists
As the economy continues to grow (Dow over 12,000), so do the non-academic opportunities for linguists. Here's an interesting one for an Analyst in the California* Bay Area :
The Analyst looks for opportunities to improve our Natural Language and Directed-dialog applications using the data logged by them. The Analyst is primarily the responsible team member charged with analyzing the data and making new implementation recommendations [...] Besides analyzing our speech applications and improving our Analytics framework, you will also have the opportunity to carry out independent research, which forms a big part of the success of our speech applications [...]
*living in the metro DC region has taught this Northern California boy that there's more than one "Bay Area."
The Analyst looks for opportunities to improve our Natural Language and Directed-dialog applications using the data logged by them. The Analyst is primarily the responsible team member charged with analyzing the data and making new implementation recommendations [...] Besides analyzing our speech applications and improving our Analytics framework, you will also have the opportunity to carry out independent research, which forms a big part of the success of our speech applications [...]
*living in the metro DC region has taught this Northern California boy that there's more than one "Bay Area."
Subscribe to:
Posts (Atom)