Literacy Project/2012-04-27 Notes

Meeting notes 2012-04-27 @ OLPC/twine

What do we need to teach?

The kids need to progress along a number of different axes simultaneously, although not all at the same rate...

Setting the Stage

Axes of Literacy I

1. Vocabulary

4 words, 20 words, 850 words (Basic English), 1,500-2,000 words

("Standard English"), 30k-40k words (fourth grader)

2. Phonemes

About 50 phonemes in English

3. Morphology (suffixes, prefixes, tenses) 4. Sentence Structure

• Subject-Verb
• Noun Phrase / Verb Phrase
• Prepositions
• Articles

5. Roots of Omo

Turns out, most of these are familiar from RAVE-O:

Axes of Literacy II

• Phonology
• Orthography
• Syntax
• Semantics
• "Und"
• Morphology

Goalposts on our Road To Literacy

Zero!                                        3rd-grade level
|------|---------//------|----------------//-----|------------|
      Unit 1            "Basic English"                   4th-grade level
      4 words           850 words                         "Charlotte's Web"
      5 letters         all 26 letters                    30k-40k words
      5 sounds          all ~50 phonemes "GPC"            This is our goal
    matching skills     most common sentence structures
    abstract matching   all roots of omo

Once they can get to the 3rd or 4th grade level, they ought to be able to bootstrap themselves up to Proust.

("Basic English" has some issues linguistically, but it's worthwhile as a vocabulary and literature source, keeping its limitations in mind.)

How can we tell if we're successful? How can we ensure that we're advancing along all the axes at appropriate rates?

Assessment Tools (and the POSSUM elements they assess)

(please help contribute good bibliographical references for the tests below, both for guidance as we implement them as well as for future scholarly discussion/writeups of the project.)

Rapid Automatized Naming (RAN)

- this is a core general technique

Peabody Picture Vocabulary Test (PPVT)

- [Se]mantics / Vocabulary

- name the picture (more details below)

Woodcock Reading Mastery Test, visual/auditory portion

- I think this is the basis for the abstract matching test?

"Word Karate"

- [M]orphological assessment

- first level, split the affixes from the root (visual)

- second level morphological assessment is matching: hear one form, select another. ie, show/speak "bat" and select one of "bats", "mats", "rats"

[CSA Q: can we introduce this in unit 2?]

LIPS - Lindamood & Bell

- [P]honology

- Show "m" or say "mmmm", select the matching face (lips closed, not mouth open), etc.

- More advanced variant -- show a picture of a face, encourage kids to match it while making the correct sound. (A bit of a research project.)

Matching test variants

show/speak "hut", choose among object pictures. [Se]mantics/Vocabulary.

show/speak "huts", choose among object pictures. [M]orphology

show/speak "the hut" or "a hut", choose among pictures. [Sy]ntax.

show/speak "the/a red hut", choose among pictures of diff objects. [Sy]ntax

show/speak "the/a red hut", choose among red hut, green hut, blue hut. [Se]mantics, Noun Phrases

show/speak "the/a hut is red", choose among hut/bat/cat. [Sy]ntax

show/speak "the/a hut is red", choose among red hut/green hut/blue hut. [Se]mantics, Sentence structure.

LIPS variant from above. [P]honology

show/speak "bat", select among words bats/mats/rats. [M]orphology

Whenever possible, supplement a correctly matched choice with feedback in a different form. For example, if we show the word "mat" and they correctly select the picture of the mat, we should reward them by speaking aloud "mat" and ideally breaking it down "em ay tee, mat" "mmm aaah t, mmmaaaht, mat". (Blending)

[CSA Q: "em ay tee" or "mmm aaah t" or both in unit 1?]

Whenever possible, accompany spoken text with a video of a human speaking the text, concentrating on the lips. This helps reinforce proper sound production. ([P]honology)

Teaching materials

We're concentrating on assessment, but the assessment tool will do some amount of teaching as a side effect. Teaching tools include:

• "Minute stories" -- read aloud, with word highlighting

- if we can show lips of speaker, that would be good

- maybe we can make these more interesting by making them more like silent films, with some limited text, so that the meat of the plot (which is hard to express with a limited vocabulary) can be acted out, and doesn't need to be textual.

- RAVE-O also gets good mileage out of using stock characters. We know Sam Sleuth is out sleuthing after bad guys, even if the text of the story doesn't say so explicitly.

- Talk to David Rose (CAST, Center for applied special technology) about using the "Book Builder" materials as a source for stories

- Stories which have been translated into Basic English

- Early American primers, from Project Gutenberg

- C. Scott hopes to have an application demo for storytelling in the near term

• "training" screens in matching tool (see below)
• Reinforcement on correct matching. (ie, when you correctly match the word with the picture, you *hear* the word as well)

Sidebar on letter shapes and # of choices

We're going to concentrate on *lowercase* letter forms. Delay introduction of uppercase as long as possible.

The matching app could have 3 or 4 options to choose from. Three keeps complexity and child-memory requirements down, but leads to greater false positives when evaluating the results. Decision: randomly choose between 3 and 4 options shown when playing the matching game, so we can crunch the numbers and see if it makes a difference. (Alternatively, each child gets either 3 or 4 options consistently; alternatively, 3 options at one site, 4 at the other.)

Sidebar on deployment

We have a three week pipeline:

• Have something new ready to go
• 1 week to test and work out deployment instructions/timing
• 1 week for the new thing to be in the hands of the kids
• 1 week for the data to get shipped back to us

Try to keep the pipeline full.

Sidebar on teaching philosophy

Three axes:

1. "Child choice" -- permissive/restrictive in what activities child is allowed to do (there's a continuum)
2. Immersion/curation/systematic -- degree to which all the materials on the tablet are carefully selected to conform to a pedagogical theory
3. Guidance ("narrative") -- degree to which a path through the material is exposed.

These axes interact to some degree, although you can imagine interesting points at the extremes: a permissive system allowing the child to do anything while still exposing a clear "best path" through the material ("the next lesson is #4, although you could click on any lesson you like"), or a restrictive system which requires you to complete "reading lesson 1" before you can attempt "reading lesson 2" while still allowing free choice among the materials which are not "reading lessons". Either of these systems could be "curated" with only relevant material, or "immersive" with a broad range of material outside the scope of the pedagogical theory or guidance.

Scott notes that we have a permissive immersive environment currently in the field, so we might as well begin with what we have and wait to remove material/restrict choices until we have data showing it's necessary (or until we can do a proper contrastive experiment).

Chris notes that discussions of "Tablet 2" threaten to push it into the unattainable future, so that it's more productive to concentrate on particular steps to incrementally transform our Tablet 1.

===================
NOW THE GOOD STUFF!
===================

Contents of "Unit 1"

1. Establishing "matching"

Color -> Color (show blue swatch, pick from red/blue/purple)

Object -> Object (show picture of cat, pick from pictures of cat/bat/mat)

Letter -> Letter (show "a", pick from a/b/c)

2. "Abstract matching"

Square -> Red, Circle -> Blue, Star -> Purple

show b&w square, pick from red/blue/purple

also/alternatively, sounds:

"sproing" -> Red, "thunk" -> Blue, "sad trombone" -> Purple

play "sproing", pick from red/blue/purple

This test is somewhat controversial -- it might be too hard, or confusing. But it sets up the letter/sound tests, and it's worth collecting the data to see whether success/failure at #3 correlates with success/failure with #2.

This test needs a "teaching screen" in the assessment app. For example: first, show a red square. then, clear the screen and show a black and white square, and prompt them to choose between red/blue/purple.

The "memory distance" between the teaching and assessment steps can be varied. Maryanne suggests starting with the teaching illustration present on the assessment screen itself. Then you can later move to showing them on separate screens, then you could decrease the frequency at which you show the teaching screens, etc.

3. Teach the five lowercase letters a/b/c/m/t.

[c will always make the k sound. letters will always be lowercase.]

3a) Sound to letter. Play "aah" or "aye". Pick from a/b/c.

[CSA Q: "aah" or "aye" or both?]

3b) Letter to sound. Show the letter. Buttons now correspond to different sounds.

CJB and CSA will work out how this interface works technically. (CSA suggests that the buttons light up one by one in a loop, saying their sound as they light up.)

3c) [CSA addition] Sound/letter to face animation.

Show "a" or say "aye" or "aah", pick from face with mouth open making A sound, face with mouth closed saying M, or face with lips making B sound.

[CSA Q: "aah" or "aye" or both? Can we get face image guidance from LIPS?]

4. Teach the four CVC words cat/bat/mat/cab.

Say/show the word, pick the picture.

BAT/CAB are MIM words (many interesting meanings!).

[CSA Q: just show 1 picture for each in unit 1, or do we want to introduce MIMs immediately?]

These five letters and four words were chosen using the following criteria:

a) build on alphabet song, so teach "A B C".

b) M and T are common letters which combine with ABC to make useful words "ETAOIN SHRDLU" (actually ETAOINSHRDLCUMWF) is not just the name of a character in Elmer Rice's "The Adding Machine", but also the frequency of letters in English.

([1])

T,A are #2 and #3 in frequency. E is ruled out based on phonological difficulty. C,M are #12 and #14; based on frequency one might consider replacing M with N (#6), which would mean replacing MAT with CAN or TAN. (In future units one might then add ANN or ANA, ANT, BAN and/or NAB.)

[CSA Q: replace MAT with CAN?]

c) BAT and CAB are MIM words, which is useful for the future.

Who's Doing What, When

Next meeting: 2pm-4pm Monday May 7 at OLPC/Twine.

• Chris and Scott will show an assessment app doing at least objective #1 from unit 1, and hopefully all four objectives.
• Maryanne, Stephanie, Denise will propose 'Unit 2' objectives.
• Fardad will contribute some "reward" media.
• Stephanie's going to send Maryanne/Cynthia's whitepaper on the literacy project to the literacy@ list for Denise, Scott and Chris.
• Richard will leave roughly Thur May 10 for Ethiopia.

Hopefully we can approve the initial assessment app on May 7 and squeeze the deployment/testing work into the week of May 7 so that Richard can supervise its deployment in the field, but the primary goal of Richard's trip is really solving the power/date issues.