Tuesday, May 27, 2008

Boy, interrupted. (by teacher)

Krashen and Brown recently published this research paper in the Singapore Tertiary English Teachers Society journal: What is Academic Language Proficiency? [download PDF]
Their paper flies in the face of current popular pedagogical 'wisdom' that suggests that nearly everything needs to be taught explicitly to students.

Specifically, they suggest that metacognitive strategies (supposedly designed to assist students with creating a deeper understanding of material) may actually be getting in the way of students' learning.  They quote the experience of one middle school teacher who had encouraged students to pause at intervals during their reading to create visual associations:
After a few weeks, her students rebelled, and told her that "Metacognition was interfering with the reading zone ... (it) disrupted the flow of a great story; ate up precious hours that could have been devoted to living inside another great story, and wasted their time as readers ... not one student could name a positive effect of the strategies on his or her reading performance".
'Conventional wisdom' usually looks at tools that have the potential to be useful, and advance the implementation of these tools or strategies in day-to-day contexts.  However, 'common sense' (which is possibly closer to 'enduring wisdom') might be worth considering when 'useful' tools and strategies are actually creating unnecessary detours from the simple enjoyment of learning.  Sometimes spending too much time on the scaffolding may unnecessarily slow down the building process.
...some strategies are teachable and useful to learn.  Others are less useful, limited only to conscious language learning and deliberate memorization.  Still others, those that all humans naturally possess and use, may be counterproductive to teach.
Ref:  Krashen and Brown (2007).  STETS Language and Communication Review, Singapore Tertiary English Teachers Society.

Thursday, April 24, 2008

Cognitive Machines

The scope of projects that are emerging from the Cognitive Machines group at MIT's Media Lab is fascinating and inspiring. Current projects include the Human Speechome Project, which is possibly the most complex longitudinal study of language development that's ever been undertaken: the language and behavior patterns of a single child have been recorded continuously in several hundred thousand hours of video and speech recordings.

Recently I've encountered questions about the viability of "The Semantic Web" (Web 3.0) on forums such as Internet Evolution, and a question has been raised as to how rich content without accompanying metadata can be catalogued and contextually searched on the web. In projects like, "Situated Natural Language Processing for Sports Video" I think we see the tip of the iceberg, in what Michael Fleischman and Deb Roy categorize as, "exploiting aspects of the non-linguistic context, or situation, conveyed by the accompanying video."

I highly recommend MIT's Cognitive Machines arm of the MIT Media Lab to anyone who's interested in emerging linguistic and sociological analysis, and the development of artificial intelligence technologies.

Wednesday, April 09, 2008

Corpus Linguistics for the Cloud Crowd


Recently cloud/cluster representations of online data have been emerging in popularity, and they're now progressing from a wider 'art form' to personally-accessible utilities. The innovative work of Jonathan Harris has produced personally-accessible such as Wordcount, an evolving graphic representation of lexical frequency in the English language. Even more accessible and playful is Tweet Clouds by John Krutsch and Jared Stein, which quickly grabs public 'tweets' from Twitter streams like mine, and generates a graphic representation of word frequency.