Language Log

Yesterday afternoon at an ICASSP-2014 session on music transcription. Just as in the session on Diarization that I wrote about yesterday, most of the papers reported results on published data, and several also offered links to their code. Thus Ken O'Hanlon and Mark D. Plumbley, "Polyphonic Piano Transcription using Non-negative matrix factorisation and group sparsity", which ends with this note:

6. REPRODUCIBLE RESEARCH
This research benefits from the efforts of other researchers to share their code [5] and dataset [21]. The open availability of these resources is commendable, allowing other researchers to easily and accurately compare methods. The code used in the experiments described in this paper is available at http://code.soundsoftware.ac.uk/projects/gs bnmf/.

From Anthony Chemero, Radical Embodied Cognitive Science:

Imagine the scene: An academic conference. Two cognitive scientists, casual but friendly acquaintances, are chatting in a hotel bar.'

"So, what are you working on now?"

"I've been doing some stuff with [insert one of: ecological psychology, connectionist networks, dynamical modeling, embodied cognition, situated robotics, etc.]."

"But [insert name(s) here] already showed that that approach is hopeless. The paper was published in …"

"Yeah, yeah. I've read that one. I don't buy it at all. [Reinsert name(s) here] doesn't really get it. You see …"

If you're reading this, you've probably taken part in a conversation like this. In fact, nearly everyone working in cognitive science is working on an approach that someone else has shown to be hopeless, usually by an argument that is more or less purely philosophical. This is especially true of the not quite mainstream approaches listed above, the approaches that constitute the core of radical embodied cognitive science, the view I will describe and defend in this book. But it is also true for more mainstream computational cognitive science (e.g., Miller, Galanter, and Pribram 1960). We all know about the arguments that purport to show that our research can never succeed; indeed, nearly every book written by a philosopher begins with an argument that the competing approaches are hopeless. Yet, for some reason, we persist. Somehow we're only convinced by the philosophical arguments that everyone else's approaches are hopeless.

Perhaps the most widespread of all Chinglish expressions (it has become virtually a standard throughout China) is "slip carefully", with extensions such as "carefully slip and fall down", "please slip carefully", and so forth.

TEXTS

IMAGES

This ubiquitous regret has long provided fodder for commercial exploitation—since the 1960s, speed reading courses have been marketed with promises to double or quadruple the rates at which text can be absorbed and understood. More recently, a number of speed reading apps for mobile devices have been released (e.g. Velocity, QuickReader, Acceleread), promising “superhuman” reading speeds that will “accelerate your learning potential” and help you “keep up with the web, blogs, twitter and e-mail.” (Now, if they could only invent an app for quadrupling the speed of answering e-mails, or writing Language Log posts, or thinking about what I’ve just read, that would truly increase my productivity.)

Of all these apps, the recently launched product Spritz offers the most specific pseudo-scientific hype as part of its marketing (their website offers a page soberly titled “The Science”). Since the stated rationalization for the app sounds plausible, at least to the point of managing to impress a number of smart, generally skeptical people who’ve sent me queries about it, it’s worth subjecting it to some psycholinguistic scrutiny.

I spent this morning at at ICASSP-2014 session on "Speaker Diarization". As the picture indicates, the room was not exactly handicapped accessible…

Luckily this is not a problem for me, but my experience of three torn knee ligaments a few years ago sticks with me.

Anyhow, I made it up the stairway to Room Scherma, and learned some useful and interesting things about current techniques for speaker diarization, which is the problem of determining who spoke when in an arbitrary audio or video recording.

At the site of ICASSP 2014 to register yesterday evening, this is what I saw:

Erratic in time, anyhow — maybe no more erratic than usual in terms of content.

Sent in by A.C. from NZ:

My ISP's sign-on page has a 'daily picture', accompanied by some surprising(?) trivia. (Usually the surprise is how strained is the link to the picture and how badly they twist the language — often ending up misusing language in some way or other — this one is itself an example: is it really a fact *about* snow?)

Even a simple Google search offers an Inuktitut word for hello. (Admittedly also a large number of links alleging "no word for hello".)

Sparing you the screenshot, the "Daily Fact" is "Snow: Eskimos have hundreds of words for snow but none for hello."

Arif Dirlik called my attention to a wonderful article entitled "'Speak Tibetan, Stupid': Concepts of Pure Tibetan & the Politics of Belonging" in the Lhakar Diaries.

At the heart of the article is this powerful 16-minute video entitled "Linguistic Diversity on the Tibetan Plateau":

A recent xkcd:

Morse code was first used for telegraphy in 1844. I'm not sure when the French navy started to use it, but I understand that transmission of Morse code by radio began in the 1890s, so if that last plaintive message was sent in 1997, Morse code in the French navy would have had a run of just about 100 years. LiveJournal was started in April of 1999, and of course some people still use it, but it flowered and faded in the U.S. over a period of about a decade: "In January 2009 LiveJournal laid off some employees and moved product development and design functions to Russia", according to Wikipedia.

Yesterday I attended a workshop on Social Media Data at the Annenberg Public Policy Center, and it occurred to me to wonder what the social media landscape will be like ten years from now.

Over at Lingua Franca, fellow Language Log author Geoffrey Pullum has an excellent article entitled "There Was No Committee".

Here's a key paragraph:

Some people talk as if Mandarin Chinese was gaining on English. It is not, and it never will. A Tamil-speaking computer scientist explaining an algorithm to a Hungarian scientist at a Japanese-organized scientific meeting in Thailand calls on English, not Chinese. Nowhere in the world do we find significant numbers of non-Chinese speakers choosing Mandarin as the medium for bridging language gaps. There are no signs of that changing.

Paul Krugman, "Why Economics Failed", 5/1/2014 (emphasis added):

Meanwhile, powerful political factions find that bad economic analysis serves their objectives. Most obviously, people whose real goal is dismantling the social safety net have found promoting deficit panic an effective way to push their agenda. And such people have been aided and abetted by what I’ve come to think of as the trahison des nerds — the willingness of some economists to come up with analyses that tell powerful people what they want to hear, whether it’s that slashing government spending is actually expansionary, because of confidence, or that government debt somehow has dire effects on economic growth even if interest rates stay low.

In Turkey, outspoken newspaper columnist Önder Aytaç has received a 10-month jail sentence over an errant "k" on Twitter.

Here is how the situation is explained in Zeynep Tufekci's widely cited Medium post:

Meet “k”, the character that got newspaper columnist and academic Önder Aytaç a 10 month jail sentence in Turkey. Aytaç is a columnist for a newspaper affiliated with the Gulenist movement, followers of Fettulah Gulen, the self-exiled cleric who lives in Pennsylvania and was once the AKP government’s closest ally, but now is among its bitterest enemies. The fight between the former allies surfaced over the closing of “private schools,” or “dershaneler,” which the Gulen movement operates in dozens of countries around the world, including the United States. These dershaneler are crucial to the movement as they are the source of both recruits and money. The Prime Minister of Turkey, Erdogan, announced in late 2013 that he would be shutting them down.

During the bitter fight, Onder Aytaç tweeted this:

http://t.co/zYBEmYiD =>> KAPAT BE USTAAAAAAMMMMMMK :) => yorumkariniz da yazinin altina lutfen

— M. ÖNDER AYTAÇ (PhD) (@onderaytac) September 20, 2012