lain tal aot literac and learnin - wordpress.com · lain tal aot literac and learnin (504) 840-9786...
TRANSCRIPT
Language at the Speed of Sight Mark Seidenberg
MARCH 13-15, 2017HILTON NEW ORLEANS RIVERSIDE
PLAIN TALK ABOUT LITERACY AND LEARNING
(504) 840-9786 | [email protected] | www.cdl.org
About the Presenter
www.cdl.org | [email protected] | (504) 840-9786
About CDLWhat does CDL do when we aren’t doing Plain Talk? Plenty.
We provide real-time, customized professional learning services that are sustained, collaborative, student-focused, and data-driven.
Nothing canned. Nothing scripted. Our professional learning is designed, facilitated, evaluated, and adjusted to meet the needs of your educators.
With a boots-on-the-ground approach, we provide collaborative learning sessions, coaching, modeling, and observations with feedback.
We examine student and teacher data with your leadership team, and then build professional learning in response to student and teacher needs.
We tackle real-time issues such as critical thinking, remediating struggling readers, and building and sustaining collective capacity.
We focus on the knowledge and skills to enable students become proficient learners who succeed in core academic subjects and meet challenging standards.
CDL’s professional learning services are relevant to the needs of your students and your teachers, and aligned with the professional learning definition in the 2016 Every Student Succeeds Act.
We have robust expertise in literacy, evidence-based strategies, how students learn, early childhood, student-specific intervention and remediation, leadership, and building collective capacity.
We have experts at every level from early childhood through high school ready to work with your educators.
Give us a call - we are ready to travel to you.
Customized.
Pragmatic.
Collaborative.
Real-time.
Focused.
Relevant and aligned.
Robust.
Diversified.
We travel.
Mark Seidenberg Mark Seidenberg, Ph.D., is a cognitive scientist/neuroscientist/psycholinguist who has studied language, reading and dyslexia since the disco era. He is the Vilas Research Professor and Donald O. Hebb Professor in the department of psychology at the University of Wisconsin. His reading research addresses the nature of skilled reading, how children learn to read, dyslexia, and the brain bases of reading, using the tools of modern cognitive neuroscience: behavioral experiments, computational models, and neuroimaging. Mark’s language research addresses what people know when they know a language, how this knowledge is represented in the brain, and how it is acquired and used. His book, Language at the Speed of Sight: How We Read, Why So Many Can’t, and What Can Be Done About It was published to acclaim and derision in January 2017. Mark attended Columbia University
as an undergraduate, where, like many students, he worked part-time. He received a Ph.D. from Columbia, where he was a student of Tom Bever during the notorious Nim Chimpsky era. He did postdoctoral research at the Center for the Study of Reading splitting time between Bolt Beranek & Newman and the University of Illinois. His first academic appointment was at McGill University in Montreal, home of fantastic food and a very long winter. After 10 years in the cold, he moved to the University of Southern California, where he had appointments in psychology, linguistics, and in the neuroscience program. In 2001 he moved to the University of Wisconsin Madison, where he lives in a house on a hill and bikes to work when he can. He has published many scientific articles in fine journals such as Science, Psychological Review, Nature Neuroscience, Language, Psychological Science, and Semiotica, and was honored as one of the 250 most-cited researchers in the areas of psychology and psychiatry by those Web of Science citation-counting people.
Center for Development and Learning(504) 840-9786 | www.cdl.org | [email protected]
Language at the Speed of Sight
Mark Seidenberg
Vilas Research Professor
University of Wisconsin-‐Madison
What can reading science contribute to be@er reading?
2/26/17 Mark Seidenberg 1 2/26/17 Mark Seidenberg 2
"Every teacher of young children as well as those who train them should read this book.”
Wall Street Journal, 1/10/2017.
On sale here!
My basic assumpRon
2/26/17 Mark Seidenberg 3
Too much to learn, too li0le 1me
2/26/17 Mark Seidenberg 4
Beginning reader’s task
2/26/17 Mark Seidenberg 5
Learn how print relates to spoken language they already know (Gough)
But now we know just how deeply they become intertwined.
2/26/17 Mark Seidenberg 6
Rueckl et al. PNAS 2014
Shankweiler et al.: integra:on greater for good readers compared to poor readers
PLAIN TALK ABOUT LITERACY AND LEARNINGNew Orleans, LA | March 13-15, 2017
1
Center for Development and Learning(504) 840-9786 | www.cdl.org | [email protected]
2/26/17 Mark Seidenberg 7
Reading and speech are “Like two co-‐dependents with serious boundary issues.” What the child has to learn
• spoken language • print: orthographic structure • relaRons between print and spoken language
2/26/17 Mark Seidenberg 8
You can’t teach all of this Too much to learn, too li0le 1me
• Vocabulary • Phonics • Spelling
(Concepts, grammar, things about the world, much more)
2/26/17 Mark Seidenberg 9
The Learning Puzzle
• Readers do learn all this • There isn’t Rme to teach all of it • Teaching a relaRvely small amount of it is beneficial
anyway
2/26/17 Mark Seidenberg 10
What is going on?
To explain this, let look at something everyone just loves to discuss…
Phonics!
Bear with me! It’s a useful example Same story for vocabulary, morphology, syntax, concepts, etc.
2/26/17 Mark Seidenberg 11
One view
Why bother?
Not good
2/26/17 Mark Seidenberg 12
PLAIN TALK ABOUT LITERACY AND LEARNINGNew Orleans, LA | March 13-15, 2017
2
Center for Development and Learning(504) 840-9786 | www.cdl.org | [email protected]
Another view
• There are rules Mat pat sat hat generalize: nat
• There are excepRons Have said done was aisle
Therefore • Teach the rules, memorize the excepRons (“sight
words”)
2/26/17 Mark Seidenberg 13
Problem
1. No one knows what the rules are 2. You can’t say which words are excepRons without
knowing what the rules are 3. See (1)
2/26/17 Mark Seidenberg 14
2/26/17 Mark Seidenberg 15
“Words in the COOK/LOOK/BOOK neighborhood seem to obey a rule governing the pronunciation of-OOK, which makes SPOOK an exception.
But SPOOK is fine if grouped with SPOON and SPOOL.
BOOK is rule governed if grouped with LOOK and COOK but irregular compared to BOON, BOOM, and BOOT.
Which -OWN words are rule governed and which are “exceptions”: CLOWN, TOWN, and FROWN or OWN, FLOWN, and BLOWN?
Worse, however they are stated, the number of rules is so large that only a small subset can be taught, surely the bane of every teacher charged with providing phonics instruction. How the child catches on to the rest is a
From the book:
What’s the answer?
• The system isn’t rule-‐governed. • It’s probabilisRc
“Rule” means: given X, do Y • Like rules for moving pieces in chess
“ProbabilisRc” means: given X, certain probabiliRes of doing X or Y or Z.
• Like wri@en English
StaRsRcal tendencies, not rules and excepRons
2/26/17 Mark Seidenberg 16
Degrees of consistency
• ”rule-‐governed” words have excepRons • But the excepRons are not arbitrary
HAVE is not pronounced “glorp”
It overlaps with “rule-‐governed words” Had, has, hive, etc.
• There is no line between rule-‐governed and sight words. Just degrees of consistency. ConRnuum.
2/26/17 Mark Seidenberg 17
How is this learned?
• “Deep learning” network • Given a spelling pa@ern,
produce the correct pronunciaRon
• Do that for every spelling pa@ern
• Use the same network for all words
2/26/17 Mark Seidenberg 18
PLAIN TALK ABOUT LITERACY AND LEARNINGNew Orleans, LA | March 13-15, 2017
3
Center for Development and Learning(504) 840-9786 | www.cdl.org | [email protected]
What happens
StaRsRcal learning procedure • The network discovers regulariRes • Lots of pa@erns, subpa@erns, weird pa@erns not
visible to naked eye • Eventually converges on configuraRon that allows it
to produce correct pronunciaRons for many words • Regulars, excepRons and everything in-‐between
• And generalize to novel items
2/26/17 Mark Seidenberg 19
Involves two types of learning
• Implicit, unsupervised • Goes on all the Rme in the background • Network/brain learns from experience ”on its own”
• Explicit, supervised • Feedback about whether behavior is correct or not • Like, instrucRon
2/26/17 Mark Seidenberg 20
Important finding
• Learning in such systems is most effecRve when it combines
• Large amount of implicit learning • Smaller amount of explicit instrucRon
• We think this is true of children
2/26/17 Mark Seidenberg 21
ExplanaRon
• What is learned about one word carries over toother words
• Learning about SAVE makes it easier to learn PAVE and RAVE
• Facilitated by well-‐aimed explicit instrucRon Doesn’t just help the child learn that specific example.It influences all of the overlapping ones.
2/26/17 Mark Seidenberg 22
Same thing for vocabulary
• StaRsRcal learning over the contexts in which words occur
• Words with similar meanings appear in similar linguisRc contexts
• Learning about lion, Rger prepares the way for lynx
• Before the child experiences it!
2/26/17 Mark Seidenberg 23
We think this is the soluRon to the “too much to learn, too li@le Rme” problem
• Don’t have to teach it all Lots of learning goes on as the child reads, writes, talks, listens
• But instrucRon is also importantTunes the system in ways that help many words at a Rme
2/26/17 Mark Seidenberg 24
PLAIN TALK ABOUT LITERACY AND LEARNINGNew Orleans, LA | March 13-15, 2017
4
Center for Development and Learning(504) 840-9786 | www.cdl.org | [email protected]
What’s going on in the child’s head is not apparent from observaRon We do the research
Behavioral, neuroimaging studies of children, adults ComputaRonal models of learning
Explicit learning: visible Rp of the iceberg Implicit learning: mass hidden below the surface
2/26/17 Mark Seidenberg 25
Side effect
What you think you are teaching the child may be very different from what they are learning
2/26/17 Mark Seidenberg 26
What’s missing from this account?
• How to opRmize limited opportuniRes for explicit instrucRon
• What we need Given the current state of the child’s knowledge, what kind
of explicit learning trial would have the biggest overall impact on learning?
Affect the most words at a Rme.
AdapRve learning procedures. Coming very soon!
2/26/17 Mark Seidenberg 27
Would it help teachers to know this stuff? I think so, do you?
Good theory of how children learn —> good pracRces
Bad theory: harder for child, teacher
2/26/17 Mark Seidenberg 28
In my Concurrent Session @1.30 Ask me (almost) anything
Topics to be determined by the audience. Such as:
How Much Do Readers/Learners Vary?
Reading vs. Literacy: Does it ma@er?
MulRple Literacies: Good for whom?
Dyslexia: Does it exist? Yes. Is it a “desirable difficulty” (Gladwell 2013)? No.
Achievement gaps: More than poverty, opportunity?
Other topics at the intersecRon of the science of reading/language/development and educaRon. Bring yours.
2/26/17 Mark Seidenberg 29
See you there!
2/26/17 Mark Seidenberg 30
PLAIN TALK ABOUT LITERACY AND LEARNINGNew Orleans, LA | March 13-15, 2017
5