[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [xmca] Deb Roy: The birth of a word
Thanks, Peter. Someone also posted on this to the local LCHC group list, and I replied with the following after watching the video (a TED presentation):
Yes, this is pretty amazing. I was reminded of the work that Lev Manovich is doing here at UCSD on cultural analytics, trying to identify quantitative patterns in large amounts of video and image data, such as changing patterns in news programs as to how much focus there is on the presenter vs the content, the rise of digital content backgrounds, etc.
But the MIT group has taken this much further, particularly in cross-linking television content to online commentary by viewers in real time. This should be the end of the Nielsen ratings, if they weren't gone already, but its also potentially the end of the survey industry as well -- why do phone surveys of hundreds when you can get real time reactions from millions. I can see the news shows commissioning this for "spin" on major events, speeches, maybe the 2012 election. And this may be worrying, because it has an inherent tendency, esp. at the current level of the technology (re semantic analysis) to grossly over-simplify what are in fact much more complex meanings being created.
I am happy to see the work on context factors, social input and settings, in the work on language development in the home. It's Gregory Bateson meets massively parallel computing (GB did some of the first in-home filming of his daughter's first years). But in relying on very simple indices, like utterance length, it's again going to oversimplify. I don't think they can analyze at this point just how the setting and the dialogue, over more than one turn, scaffolds a sense of meaning for the child. Much easier of course to trace the growth of phonology and single word acquisition. Still it's a good step.
Quite fascinating to see something Ivan and I were predicting last year: people getting used to multi-video displays, where in this case you see simultaneous video across about 6 rooms in the house in 6 video views, and then all the tv/cable channels at once, dozens of small video displays in a giant array. How to see this? Of course their visual magic of re-rendering this into a 3D fly-through view of the whole house eliminates the simultaneity in favor of sequentiality, and some neuroscience work suggests that we are best at doing sequential pattern recognition. But even a multi-video view can appear sequential to the brain when it is visually scanned in real time by the eyes' movements and attention focusing.
Every other word he says is about privacy concerns, but you still can't disguise the Big Brother potential here: total panopticon surveillance, video and audio, 24/7 in private as well as public settings. In the UK there are already serious concerns being raised about access to the ubiquitous outdoor security cam footage, massively increased in the last 10 years everywhere in the country, as it leaks from the anti-terrorism units for whose use it was originally justified to local police departments, etc. Combining this with effective video and semantic pattern recognition algorithms presents a real danger to privacy and freedom.
Tis a good wind that blows no ill.
Senior Research Scientist
Laboratory for Comparative Human Cognition
University of California - San Diego
9500 Gilman Drive
La Jolla, California 92093-0506
Professor (Adjunct status 2009-11)
School of Education
University of Michigan
Ann Arbor, MI 48109
City University of New York
On Mar 11, 2011, at 12:30 PM, Peter Smagorinsky wrote:
> MIT researcher Deb Roy wanted to understand how his infant son learned language -- so he wired up his house with videocameras to catch every moment (with exceptions) of his son's life, then parsed 90,000 hours of home video to watch "gaaaa" slowly turn into "water." Astonishing, data-rich research with deep implications for how we learn.
> xmca mailing list
xmca mailing list