Language Log

Following on from Barbara Partee's example of Kruschev not banging his shoe, I just came across a great example of chained hypothetical negative events. It was during Bonnie Webber's plenary address here in Austin yesterday, at the NASSLLI Summer School. (BTW, if you'll be in the Austin area on Saturday, I have an announcement for you: NASSLLI is hosting a big event commemorating the centenary of Turing's birth, and it's free and open to the public.) But without more ado, here's the "Not raising hogs" text, a good Texas story of how to get something from nothing:

THE NOT RAISING HOGS BUSINESS

To: Mr. Clayton Yeutter
Secretary of Agriculture
Washington, D.C.

Today's xkcd:

Mouseover title: Although the Markov chain-style text model is still rudimentary; it recently gave me "Massachusetts Institute of America". Although I have to admit it sounds prestigious.

John Markoff, "Troves of Personal Data, Forbidden to Researchers", NYT 5/21/2012:

When scientists publish their research, they also make the underlying data available so the results can be verified by other scientists.

(I wish this were generally true…)

At least that is how the system is supposed to work. But lately social scientists have come up against an exception that is, true to its name, huge.

It is “big data,” the vast sets of information gathered by researchers at companies like Facebook, Google and Microsoft from patterns of cellphone calls, text messages and Internet clicks by millions of users around the world. Companies often refuse to make such information public, sometimes for competitive reasons and sometimes to protect customers’ privacy. But to many scientists, the practice is an invitation to bad science, secrecy and even potential fraud.

For those who don't care much about science, and oppose data publication on the basis of some combination of beliefs in corporate secrecy, personal privacy, and researchers' "sweat equity", here's a stronger argument: lack of broad access to representative data is also a recipe for bad engineering.  Or rather, it's a recipe for slow to non-existent development of workable solutions to the the technical problems of turning recorded data into useful information.

At the recent DataEDGE workwhop in Berkeley, as well as at the recent LREC 2012 conference in Istanbul, I was unpleasantly surprised by the widespread lack of awareness of this (in my opinion evident) fact.

I'm in Berkeley for the DataEDGE Conference, where I'm due to participate in a "living room chat" advertised as follows:

Size Matters: Big Data, New Vistas in the Humanities and Social Sciences
Mark Liberman, Geoffrey Nunberg, Matthew Salganik
Vast archives of digital text, speech, and video, along with new analysis technology and inexpensive computation, are the modern equivalent of the 17th-century invention of the telescope and microscope. We can now observe social and linguistic patterns in space, time, and cultural context, on a scale many orders of magnitude greater than in the recent past, and in much greater detail than before. This transforms not just the study of speech, language, and communication but fields ranging from sociology and empirical economics to education, history, and medicine — with major implications for both scholarship and technology development.

On Friday, July 20, at the 2012 meeting of the Council of Writing Program Administrators in Albuquerque NM, there will be a session called "Help Wanted: Sharing Data for Research on Reading and Writing".  Here's the proposal that was submitted for this session:

Every year since 2005, speech synthesis researchers have organized a yearly Blizzard Challenge, "[i]n order to better understand and compare research techniques in building corpus-based speech synthesizers". Part of the research effort involves the general public, who are invited to perform a series of evaluations of the results.

Participation takes about one hour in total — but your participation is registered, so that you can leave at any point, and then return and take the evaluation up again at the point where you left off. If you're willing, please follow this link to enroll and participate.

Yesterday evening at dinner, some members of the LSA Publications Committee were idly discussing the changes over time in fashions for given names.  It's obvious that things change — but it's less obvious whether these changes are cyclic. It makes sense that out-of-fashion names might come back after a generation or two — but does this really happen on a regular basis?

For the past couple of years, Google has provided automatic captioning for all YouTube videos, using a speech-recognition system similar to the one that creates transcriptions for Google Voice messages. It's certainly a boon to the deaf and hearing-impaired. But as with Google's other ventures in natural language processing (notably Google Translate), this is imperfect technology that is gradually becoming less imperfect over time. In the meantime, however, the imperfections can be quite entertaining.

The longer it is, the higher the rating:

As Wikipedia explains,

Watson is an artificial intelligence computer system capable of answering questions posed in natural language,[2] developed in IBM's DeepQA project by a research team led by principal investigator David Ferrucci. Watson was named after IBM's first president, Thomas J. Watson.

But as a page at AT&T Labs Research tells us,

AT&T WATSON^SM is AT&T's speech and language engine that integrates a variety of speech technologies, including network-based, speaker-independent automatic speech recognition (ASR), AT&T Labs Natural Voices® text-to-speech conversion, natural language understanding (which includes machine learning), and dialog management tasks.

WATSON has been used within AT&T for IVR customers, including AT&T's VoiceTone® service, for over 20 years during which time the algorithms, tools, and plug-in architecture have been refined to increase accuracy, convenience, and integration. Besides customer care IVR, AT&T WATSON^SM has been used for speech analytics, mobile voice search of multimedia data, video search, voice remote, voice mail to text, web search, and SMS.

A few days ago ("Evaluative words for wines", 4/7/2012), I illustrated how a trivial method can help us uncover the contribution of individual words to the expression of opinion in text. For this morning's Breakfast Experiment™, I'll illustrate an equally trivial approach to learning how words fit together structurally, using the same small collection of 20,888 wine reviews.

There are two basic reasons for the increased interest in "text analytics" and "sentiment analysis": In the first place, there's more and more data available to analyze; and second, the basic techniques are pretty easy.

This is not to deny the benefits of sophisticated statistical and text-processing methods. But algorithmic sophistication adds value to simple-minded baselines that are often pretty good to start with.  In particular, simple "bag of words" techniques can be surprisingly effective. I'll illustrate this point with a simple Breakfast Experiment™.

The folks at Google Labs pulled out all the stops to show off some of the Really Advanced Search capabilities that we can look forward to in coming months…