Friday, August 27, 2010

/ay/, Animated.

The Animation

Click on the image to see the animation.

The Data

The data underlying the animation consists of 928 tokens of /ay/ drawn from an interview with a 60 year old Philadelphian. The data was transcribed by an undergraduate supported by an NSF grant. It is coincidental that I was also the interviewer. A forced alignment of the transcript to the audio was performed using the Penn Phonetics Forced Aligner (P2FA). I extracted formant measurements at 6 millisecond intervals from every stressed /ay/ using Praat. I coded contextual information based on a syllabification of CMU dictionary transcriptions.

One super-long token of /ay/ was excluded because it was extremely poorly tracked, possibly due to a misalignment.

The Analysis

I rescaled the time variable to between 0 and 1 for all tokens. I then fit a smoothing spline anova model for F1 and F2 in R using ssanova() from the gss package with the following formulas
  • F1 ~ Voice*log(Duration)*Time
  • F2 ~ Voice*log(Duration)*Time
These models took a long time to fit. Using these F1 and F2 models, I got the predicted fits for F1 and F2 values at given time point in a vowel of a given duration in a given voicing context.

The Animation (again)

The "velocity" of the "gesture" is represented in two ways:
  1. The larger the point, the slower the velocity.
  2. The bluer the point, the slower the velocity.
However, these two indicators have different scales.
  1. Size: Size represents velocity relative to vowels of any duration. Two points of the same size in a short vowel and a long vowel represent the same velocity
  2. Color: Color represents velocity relative to vowels of the same duration. So, a very blue point means "short for a vowel of this duration." Points with the same color from vowels of different durations do not necessarily represent the same velocity.

The x and y axes are negative logged hertz values, and are constrained so that an inch of plot space corresponds to the same amount of negative logged formant space for both x and y.

I generated 100 frames representing a smooth transition from the minimum duration to the maximum duration. At some point, voiceless context /ay/ disappears. This is because no pre-voiceless /ay/s were longer than 0.240 seconds.

Each frame was generated using the ggplot2 library in R, then saved to a .png. Then, I used png2swf from swftools to sew the .png's together into a flash animation running at 15 frames per second.

Room for Improvement

The formant data was very messy. I simply set the maximum formant and number of formants for the entire file, without making any adjustment. I might try to implement some kind of estimate evaluation like from Keelan Evanini's dissertation, except bootstrapping from the speaker's own data.

Thursday, August 26, 2010

Dr. Laura Effed Up

For those who haven't heard, Dr. Laura Schlessinger, a well known conservative radio talk show host, has recently retired from radio after using the N-word multiple times during a conversation with an African American caller. You can read a transcript and listen to audio here if you so wish. It's just as ugly as you would imagine it to be.

Usually when a story like this comes up, there are three questions that are repeated over and over in the public discourse.
  1. What is the nature of my right to free speech vis-a-vis offensive speech?
  2. Doesn't speaker intent matter?
  3. Why can African Americans use the n-word as an in-group term, but I can't?
As far as 1. goes, I'm a linguist dammit, not the ACLU! But, no one has been arrested, for what they said, and Dr. Laura was not forced off the air by a government agency. As Word said
In the United States we can say anything we want as we are protected by the first amendment, but that does not mean that it will or should always be tolerated.
And that's all I'll have to say about that.

As for 2. and 3., I think these are not unreasonable questions to ask in a cultural vacuum. After taking cultural history into account though, the usual answer to 2. and 3. is that these are just the prices that you pay after centuries of continuous social domination, which you have (albeit unknowingly) cashed in on and benefited from in your own life. I would say it is a good exercise to accept that as a sufficient answer, although I understand why that will really piss some people off. Some bad things happen to everyone, and as Victor Frankl said
a man's suffering is similar to the behavior of gas. If a certain quantity of gas is pumped into an empty chamber, it will fill the chamber completely and evenly, no matter how big the chamber. Thus suffering completely fills the human soul and conscious mind, no matter whether the suffering is great or little.

But, the whole reason I'm making this post is that I think there are good answers to 2. and 3. even within a cultural vacuum (ish).  When news stories like these come up, I'm specifically reminded of a paper by Chris Potts on expressives.  The paper in question is called "The expressive dimension", and it appeared in Theoretical Linguistics in 2007 (you can find a copy on his website). In this paper, Potts defines expressives as words or phrases which indicate an attitude of one entity towards another, and specifies that attitude's intensity and positivity/negativity. I should also note that Potts' paper was not written to address the morality of public usage of offensive speech. Rather, he was addressing theoretical questions within pragmatics, and I have decided to apply his reasoning to this specific case.

The aspects of expressives which Potts identifies which are crucial for answering questions about 2., speaker intent, are their independence and immediacy. Potts says that expressives are independent from the propositional content (what is being said) of the sentences they're embedded in. To take his example:
(4) That bastard Kresge is famous.
(5) Kresge is famous
(4) and (5) mean the same thing in a very strict sense, but adding the expressive "that bastard" adds an additional expressive meaning. This goes to say that it is not necessary to demonstrate that Dr. Laura called some an n-word in order to claim that she entered the attitudes conveyed by the n-word into the discourse, because the expressive content of an utterance is independent from the propositional content of an utterance anyway.

Immediacy is closely related. As Potts says
the act of uttering an expressive morpheme is suffcient for conveying its content.
That is, expressives are essentially performative. By merely saying the word, Dr. Laura was immediately introducing the attitudes associated with it into the context. Just like you can't unring a bell, you can't unsay the n-word.

(Incidentally, the argumentation of the previous two paragraphs is a large part of why I've decided to only write "n-word" in even this dispassionate and theoretical blog post.)

I think reasoning from Potts' paper can also address 3., the use of the n-word as an in-group term. He says that another property of expressives is perspective dependence. Remember how I defined an expressive as "words or phrases which indicate an attitude of one entity towards another, and specifies that attitude's intensity and positivity/negativity"? Well, the n-word could be understood to indicate this attitude between these entities:
  • People who find blackness hateful...
  • ...have explosively negative attitudes towards...
  • people.
Now, if two African Americans are having a conversation, and one calls the other the n-word, the attitude which as been entered into the context is one of some third party against both of them. From there, it's not very hard to understand how the n-word became a recognition of group identity. So, to answer Dr. Laura's question, that's why black comedians can use the word, but you really really can't.

It's easy to imagine something that started off as an in-group term being used as an epithet also. Let's say that "dude" means something like
  • The Man...
  • mildly irritated by...
  • ...youth culture.
Kids would call each other dude all the time, but if a teacher said "You failed the test, dude," that would be a clear epithet. Of course, it would be no where near as awful as the n-word, because the intensity of the attitude conveyed by "dude" is much less.

Tuesday, August 24, 2010

Language Use and Aging

There was a recent scene from Louis C.K.'s new show Louie which highlights another reason why language change really bugs some people: it makes them feel old. At least, that was my reading of the scene combined with my familiarity with C.K.'s other work.

Feeling depressed after dropping his girls off with their mother for a week, Louis has a one-off smoking and drinking bender with his neighbor. He wakes up the next morning extremely hung over, and ventures out for a coffee.

Clearly Louie feels like everything the young, hip people around him are saying is both meaningless and incomprehensible.

Monday, August 23, 2010

Hyena Vocalizations

Recently, I've been helping a friend analyze some Hyena vocalizations he recorded in the field. He's specifically looking at whoops. Here's a recording of a hyena whoop that a different researcher posted to the web: Whoop (paper). It's been a lot of fun and, with the permission of the researcher, I'd like to show you a spectrogram and a pitch track of a whoop.

There, wasn't that nice?

"People’s spoken language should go unmolested"

Time for a sociolinguist in the news Shout Out, or SITNSO. There is a brief editorial piece on Kirk Hazen in the NYT today, called Say it Loud (via David Durian).

“People’s spoken language should go unmolested,” Dr. Hazen told the grateful students that day, while also urging them to embrace change. “All living language is change.”

Friday, August 20, 2010

On Marc Hauser and Science

Things are looking very bad for Marc Hauser(via Language Log), and it's been sending shivers up and down the spines of a lot of people I know. What he's been accused of seems like the worst thing that could happen. It's like learning that someone from your neighborhood is a serial killer; the stuff of hushed conversations in the hallway. "Oh my god! I was just about to cite that paper too. And I know a guy whose brother was an RA in his lab. I mean, everyone knew some of this stuff was a stretch, but this?"

I'd like to reflect on what this news means for science. Hauser's behavior was, if the final verdict ends up being what it looks like it's going to be, completely reprehensible. Fudging data is antithetical to the progress of science, and extremely damaging to public understanding of science. To quote Shaggy 2 Dope from the Insane Clown Posse:
Fucking magnets. How do they work? And I don't want to talk to a scientist. Y'all motherfuckers lying, and getting me pissed.
Insane Clown Posse, Miracles
The proliferation of attitudes like Mr. Dope's makes Hauser's (alleged) actions even more shameful, especially as it seemed like he tried to craft a lot of his work for broader public consumption.

But at the end of a day, even though a scientist failed, science was successful. According to the Chronicle of Higher Ed's story, research assistants in Hauser's lab successfully utilized the scientific method to discover inconsistencies in the data. Hauser's coding was not reproducible from the raw data, eventually leading one of his research assistants to report him to higher ups at Harvard. Here is a clear case of scientific progress at work, with no regard for the status or authority of the researchers involved.

And, I think the consequences will be positive, over all. This should serve as a reminder to us all of the importance of reproducibility or our research. There will be broader demand to demonstrate that we are not Hausering the data (you heard it here first!), which means publishing more raw data, which means cleaner, more careful research.

Thursday, August 19, 2010

Britney Spears Tongue

This is a pretty interesting thing. Apparently, when singing (or at least appearing to sing in music videos) Brittany Spears has apico-labial /l/. Look up any given video of her being interviewed, and she's not doing it.

Is this a singing / acting thing?

This live footage shows her doing the same thing.

This is quite an interesting salient non-acoustic style thing.

Monday, August 16, 2010

On the bagel lady

Briefly, I think the Starbucks bagel lady's argument is probably semantically sound, but not pragmatically. Thus, she was not engaging in good social behavior.

She ordered a whole wheat bagel, and refused to specify that she didn't want anything on it. Then, she got belligerent and was escorted out by the cops. Her argument is that
When you go to Burger King, you don't have to list the six things you don't want.
So, that actually makes good sense to me. The list of things that she doesn't want is potentially infinite, so if I had to design an efficient computational system for taking orders, it would return only exactly what was requested.

However, let's get realistic / pragmatic. Who ever wants just a bagel? I'm actually someone who frequently orders everything bagels as-is, and even if I say "I'll have an everything bagel as-is," the baristas will usually clarify "So, no cream cheese or anything? Not even sliced?" Having bagels with nothing on them is unusual, and I accept that. There is a certain convention that an order for a bagel is an order for a bagel plus some spread. That's just a fact of our social world.

This lady was upset that she was asked to specify "with nothing," but I bet she'd also be pissed if she asked someone if they knew what time it was and they replied, "Yes."

Language Use and Morality

People really, really like complaining about language use. Just check out the "peeving" tag on Language Log where they, of course, discuss the peevers rather than peeve themselves. The thing that has always struck me about peevers is how they view non-standard language use and language change through a moral lens. They certainly utilize moral language in their rhetoric. A great example is the  Queen's English Society. Some choice quotes from their website (emphasis is my own):
The Society has been concerned about the decline in standards in the use of English for many years.  Our language faces a number of challenges, as it becomes ever more widely used by people with ever less knowledge of it and respect for it.
We aim to: [...] Help our youngsters to learn English and enjoy using it properly.
Every time something goes wrong [...] we hear the phrase "communications failure." But nobody will plainly admit that it was a failure to read and write documents in standard English.
The QES could become the recognised guardian of proper English and we would strive to halt the decline in standards in its use.
The focus on downward decline, the fate of our children, and the failure to recognize the true source of our problems could all be lifted from an evangelical preacher's call to recommit our lives to Christ.

Bill Labov has also noted to us in class that language change is unique among social changes. He said that there are some older people who keep up with fashion, music and technology, but no one has ever been interviewed who said "You know, it's really great the way kids are talking these days. They're just doing great things with the language, and I hope they keep it up."

I have to wonder why language is moralized. A lot has to do with class and social structures certainly. Certain uses of language become associated with certain groups of people, and then attitudes towards those groups of people are transferred to the uses of language. But I think there might also be more to it than social attitudes. For instance, I doubt the woman who recently engaged in what amounts to linguistic (not-so) civil disobedience in a Starbucks was motivated by her social attitudes towards baristas.  Likewise for the emotional reports by some that things like misplaced apostrophes "make [them] want to stab bunnies."

I think it would be worth evaluating this peeving as being exactly what it looks like: a form of moral judgment. There was a TED talk that captured my imagination on this topic a while ago by Jonathan Haidt. He was discussing his Moral Foundation Theory. He hypothesizes that there are 5 moral foundations of moral judgment:
  1. Harm / care
  2. Fairness / reciprocity
  3. Ingroup / loyalty
  4. Authority / respect
  5. Purity / sanctity
Here is his talk:

The focus of his talk is on the difference between liberals and conservatives. Liberals tend to to focus on the harm/care and fairness/reciprocity moral foundations and reject the the aspects of ingroup/loyalty, authority/respect and purity/sanctity which have repressive social effects. Conservatives, on the other hand, embrace all five foundations given the premise that order and security are better than chaos. The crucial quote from his talk comes at 14:45.
The great conservative insight is that order is really hard to achieve, it's really precious, and it's really easy to lose.
That, I feel, must be the point of view of all peevers who take a moralistic stance on language use. In a sense they are right. The order and structure of human language is what allows it to function. Were that structure to crumble, so might many of our most useful and valuable human institutions.

However, that is where the peevers' premises are false. Language has been changing ever since humans have been speaking (by hypothesis), and there is no sense in which any observable changes in the available historical record have either contributed to or detracted from the orderliness of language. I'm not speaking from a hippy dippy anything goes social attitude either. What I mean is that in order for the Queen's English Society to even begin making the moral judgments they do, they would have to empirically demonstrate that the following conditions obtain:
  • There is a meaningful scale of measurement for the orderliness and logicalness of a particular spoken language.
  • The long chain of demonstrable language changes which took place between Proto-Indo-European and Modern English have been, on the whole, optimizing the orderliness of the spoken language.
  • The contemporary language changes currently occurring in Modern English are, on the whole, destructive to the orderliness of the spoken language.
The broad consensus of the language scientific community is that the first condition does not obtain, and therefore the following two conditions are not meaningful.

In conclusion, I feel bad for the peevers' misapprehension. Too many people already spend too much time engaged in moral outrage behavior based on false premises. These language peevers could be spending their mental time and energy on something actually socially useful. On the other hand, I am irritated by their anti-rationalism. When it comes to language peeving, there are empirical facts that are relevant to the very formulation of moral judgments, but these facts are surprisingly of little importance or interest to the peevers. That is a peeve of mine.

Tuesday, August 3, 2010

"My aunt had a psychopathic brain, and she lived to be 90."

There was recently an interesting NPR story about a neuroscientist who studies the biological basis of psychopathy. The surprising twist is that he himself has genetic and brain activity indicators of being a violent psychopath! But, he seems like such a nice guy. The conclusion of the story is that we are not simply products of our biological determinants; environmental factors also play a very strong role.

And of course, that must be true. Just look at language. We speak a particular language because we have a genetic predisposition to do so, and then we were exposed to a particular language environment. The fact that nature / nurture arguments are still raging with regards to language, which is a pretty self contained system, makes me think that they will take a long time to be resolved for issues like psychopathy.

However, I think that one of the reasons this is such an interesting story is because it's emotionally satisfying, and that's not a good thing. It's too easy to hear this story and think, "He found this brain activity is related to psychopathy, but he has that brain activity, and he's not a psychopath. Therefore, people are magic."

First off, I don't know that Dr. Fallon has figured out the risk of psychopathy associated with having this brain activity. From what I understand from this NPR story and his TED talk, he's figured out that people who are psychopaths tend to have this particular brain activity, but I don't think he's done the study to figure out how many people with this brain activity are psychopaths. That second proportion is the necessary one to know in order to evaluate how surprising or unsurprising Dr. Fallon's peaceful existence is.

Secondly, as with almost all determinants of any kind of outcome, they only operate probabilistically. For instance, I think it's pretty much accepted wisdom that smoking is highly correlated with a variety of health problems. However, everyone also knows someone, the mythical aunt from the title of this post for instance, who smoked every day and lived to be very old. That's not contrary evidence to the fact that smoking causes health problems. In fact, it's expected that some people would have no health problems, all else being equal. That's the definition of probability.

So, even if the risk of psychopathy was very high given this brain activity (which again, it might not be), the fact that some one person with this brain activity is not a psychopath shouldn't be surprising, and doesn't really call the biological basis of behavior into question. The fact that he also happens to study the biological basis if psychopathy just happens to sex the story up.

The moral of the story is to think twice about science journalism that has an emotionally satisfying end to it.

Lost: Season 6 and Linguistics

I've been watching the entire series of of Lost this summer. I've been ok with the fantastical and bullshitty stuff. For instance, CPR always works, and I don't know what an L4 vertebra is, but if you have one, you're basically fucked on this show. I'm even comfortable with the fact that not everything will make sense in the end.

However, I'm starting to get bugged how they've decided to run roughshod over my domain (linguistics) in season 6 all of a sudden! My major beefs so far are:


1) I don't care how long Richard was on that Island. There's no way he can be speaking unaccented English in 2010. When he got there, he was in his early 40s and just learning English L2. Now all of a sudden he's got a perfect American accent!? I think there's no doubt that 40 is way outside he critical period!

2) This is more nitpicky (maybe), but would an 1800s Spaniard really have said "etoy"? /s/ deletion is very common in American Spanishes, and I believe it clearly distinguishes American from Europe an Spanish nowadays.
[edit: Apparently there is s-deletion in Canarian Spanish. It probably still wasn't period appropriate, but whatever.]
 I think the show should get props for giving the actor a chance to showcase his bilingualism though (Nestor Carbonell is Cuban American). Maybe that fact un-peeves this peeve.

2 (again, cause the last one didn't count)) Jacob hits Richard in the back of the head and says in such an American accent, "Whad're you doin' here?"(that "d" represents a flap) Now, I don't know the whole deal on Jacob yet, but I get the gist that he's maybe supposed to be even ancient. Couldn't they at least go with the cinematic trope that ancient = British accent?

3) Sun runs into a tree and then forgets English? And Jack thinks it might be temporary aphasia which "affects the language center in your brain." It was nice that the writers wrote the following dialogue between Miles and Lapidus:

Miles: She hits her head and forgets English? We're supposed to buy that?

Lapidus: ...asks the man who communes with the dead.

Ok writers, I get it. Freaky stuff happens on this show. But it would be freakier if, say, she had some mystical vision, then couldn't remember English, but after working through her internal conflict over leaving the island, she gains her English back. A simple log to the head doesn't invite mystical bullshit.

Disqus for Val Systems