Thanks so much for your response Vanderson! I wish I had known about How to Teach Your Child to Read in 100 Lessons when I was teaching my daughter. I definitely fumbled my way through it :).
> I think you have an up hill battle with getting traction with DI, but if you succeed that would be a great benefit to many people. (my perception)
I completely agree and it really isn't too surprising that DI never took off in schools. FWIW, I am looking to revive the underlying pedagogy (e.g., teaching sequences for non-comparatives, joining forms, etc.) rather than DI proper.
> What is the protocol on this site for starting conversations "offline"?
Interesting question that I never stopped to consider :). I have met dozens of people through HN, and it seems congruent with the hacker ethic ("throughout writings about hackers and their work processes, a common value of community and collaboration is present"), but ::shrugs::.
Hi numeromancer, I am surprised to hear mention of Engelmann and Project Follow Through here on HN :). I would love to connect and learn more about how you discovered DI. FWIW, I am a former principal engineer at a FAANG company turned educational researcher / edtech founder. I am working to revive Engelmann's Theory of Instruction and democratize access to the explicit, systematic teaching of the skills of reading (e.g., phonological awareness, decoding, and sight recognition).
Hi Vanderson, I am surprised to hear mention of Engelmann here on HN :). I would love to connect and learn more about how you discovered DI. FWIW, I am a former principal engineer at a FAANG company turned educational researcher / edtech founder. I am working to revive Engelmann's Theory of Instruction and democratize access to the explicit, systematic teaching of the skills of reading (e.g., phonological awareness, decoding, and sight recognition).
In 2018, I took a year-long sabbatical from a FAANG company (FYI, I was principal level and it required VP approval). I had previously taken a few 1-3 month breaks between positions prior to the sabbatical. Although am I grateful for the extended breaks, they are far from a panacea and downright stressful if you set your expectations too high :).
Feel free to reach out if you want to discuss further.
> Framing education as lessons presumes that lessons are how people learn
To clarify, are you thinking of a lecture instead of a lesson? A lecture is "a speech read or delivered before an audience or class, especially for instruction or to set forth some subject" whereas a lesson is "a section into which a course of study is divided".
A lesson is a generic term agnostic to instructional style (e.g., explicit, discovery, project-based, etc.) Similar to how everything in software engineering is a "task" :).
FWIW, this framing is the start of an essay I am writing about why the untold number of apps that purport to teach children to read fall far short and what can be done about it.
It depends how far down the rabbit hole you want to go :). I highly recommend checking out APM's documentaries on this issue (https://features.apmreports.org/reading/). These are in-depth and accessible.
If you want to go further, you can read Moat's Speech to Print, Seidenberg's Language at the Speed of Sight, and many others. If you want to go even deeper, then welcome to the firehose that is educational research :D.
> surely the literacy crisis is a socio-economic problem not a technological one.
Yes and no. It is, of course, not strictly a technological one, but the argument that it is a socio-economic one is, at best, an oversimplification. If you are interested in a more complete understanding, I highly recommend checking out APM's documentaries on this issue (https://features.apmreports.org/reading/).
From my research, the underlying causes of the literacy crisis are:
1. The mistaken belief that reading, like speaking, is biologically natural. This belief manifests as guidance to surround your child with books and read to them. Unfortunately, this isn't sufficient for the majority of children.
2. The majority of teachers lack the content knowledge to teach children to read. For example, imagine helping a child to sound out the word "father". What is the sound of the second letter? It isn't a short 'a' nor a long 'a'.
3. Many popular programs used in schools are completely debunked by science (e.g., cueing theory), but as a teacher it is difficult to identify that your approach is faulty. (If ~30% of children learn regardless of method, it is too easy to offer excuses for why the other children don't learn).
4. Helping a struggling child is "rich man's game". If you are high SES and your child is struggling, you will pay a tutor to rectify the problem. That isn't an option for the vast majority of families.
In other words, this is a highly complex puzzle and it is completely understandable why society is seemingly no closer to solving it :). Consequently, the majority of our effort is directed at understanding these root causes and identifying how to overcome them (FWIW, we have made significant progress here). The cognitive modeling portion is a small but plausibly important part of the larger landscape.
So far, I am a week into learning ML :). I have spent ~30 hours watching various ML courses and am in the process of testing the hypothesis that teaching reading with a shallower orthography (e.g., differentiating between the short and long 'e' sounds by introducing an 'ē' grapheme) leads to improved recognition of sublexical patterns. The step I am working on is building an embedding layer to ensure that these new graphemes (i.e., 'ē', 'ā', etc.) are near their parent grapheme (i.e., 'e', 'a') in the embedding space. (Although the model seems straightforward, I could also be completely misguided in how I am tackling this problem :) ).
FYI, this orthographic approach (i.e., how words are spelled using an alphabet) is used in a few highly researched literacy programs, but AFAICT there isn't direct research on the approach itself. The motivation is to initially make English a consistent language (i.e., the letters you see have a one-to-one correspondence with a particular sound). This should greatly simplify the initial roadblock in learning to read English (as seen by studies of countries with "shallow" orthographic languages) and then learners would transfer this knowledge to the normal (inconsistent) English orthography.
My main goal is to use cognitive modeling to evaluate the efficacy of interventions and inform the personalized "minimum effective dose" for a particular learner. Academically, this is well-trodden territory [0-2] but these results haven't found there way into practice. This is critically important because we know that ~30% of children will learn to read regardless of method, ~50% require explicit, systematic instruction, ~15% require prolonged explicit and systematic instruction, and up to 6% have severe cognitive impairments that make acquiring reading skills extremely difficult [3]. Yet, how much is enough?
To make this more concrete, imagine you are learning a foreign language with Duolingo. How much effort per day is necessary to achieve that? Many people have long streaks and are no closer to fluency (I learned nearly nothing despite a 400 day streak). Similarly, many reading interventions are once-a-week and, predictably, don't meaningfully affect the learning outcomes for those students.
BTW, this ML portion is part of a much larger effort (e.g., our team is a Phase II finalist in the Learning Engineering Tools Competition). If anyone is interested in collaborating, please feel free to reach out to me.
[3] Education Advisory Board. (2019). Narrowing the Third-grade Reading Gap: Embracing the Science of Reading, District Leadership Forum: Research briefing
I seem to be in a similar situation as an experienced software engineer who has jumped into the deep end of ML. It seems most resources either abstract away too much detail or too little. For example, building a toy example that just calls gensim.word2vec doesn't help me transfer that knowledge to other use cases. Yet on the other extreme, most research papers are impenetrable walls of math that obscure the forest for the trees.
Thus far, I would also recommend Andrej Karpathy's Zero to Hero course (https://karpathy.ai/zero-to-hero.html). He assumes a high level of programming knowledge but demystifies the ML side.
--
P.S.
If anyone is, by chance, interested in helping chip away at the literacy crisis (e.g., 40% of US 4th graders can't read even at a basic level), I would love to find a collaborator for evaluating the practical application of results from the ML fields of cognitive modeling and machine teaching. These seemingly simple ML models offer powerful insight into the neural basis for learning but are explained in the most obtuse ways.
Congratulations on launching! I love the combination of an app coupled with physical books. It really demonstrates your understanding of your customers.
* I would love to know more about your phonics instruction. From what I can see, Ello embeds the instruction into the "help the child sound out the words" component. Is that correct or is more systematic instruction provided?
* How do you help children overcome the seemingly maddening inconsistency of the English language? For example, the long a sound can correspond to the a, ai, eigh, aigh, ay, er (RP), et, ei, au, a_e, ea, and ey graphemes.
* Does your speech recognition system understand at the phoneme level?
BTW, I would love to connect. I am an independent educational researcher and early literacy edtech founder. My background is big tech (principal engineer at a FAANG company) and am now navigating the treacherous waters of building an audience, securing funding, bringing in experts, etc.
We are building an early childhood literacy app focused on explicitly and systematically teaching the skills of reading (e.g., phonological awareness, decoding, and sight recognition.) The mission is to democratize access to evidence-based instruction and raise the percentage of proficient US 4th grade readers from 33% [1] to an estimated 94% [2][3].
If anyone is interested in collaborating, please reach out. I am especially looking for experts in sales and marketing.
[2] Education Advisory Board. (2019). Narrowing the Third-grade Reading Gap: Embracing the Science of Reading, District Leadership Forum: Research briefing
[3] Although the estimates of reading capability are for children to reach NAEP’s Basic level, the mission of this project is for every child to reach NAEP’s Proficient, which requires more than just the skills of reading.
I truly struggled with writing for the longest time, but three practices helped tremendously. The practices are 1) writing with a piecemeal approach, 2) accumulating notes as fuel for tomorrow’s writing, and 3) rewriting with keyword outlines.
My most important shift was to start writing nonlinearly. Instead of writing a beginning, middle, and end, I instead gather together all the claims, facts, etc. and develop them individually without concern for the overall logical structure. Eventually, the pieces start fitting together and the linear structure emerges. It is so much easier to start with too much and whittle it down.
Where do these claims, facts, etc. come from? This is what your note taking system should create. As I read a text, I highlight relevant sections and then go back through to paraphrase them. These notes are organized around topics and solve the daunting blank page problem.
How should you paraphrase? Keyword outlining is the practice of picking a handful of _keywords_ from a source text, setting aside the original text, and then paraphrasing using the keyword outline and your recollection of the original text. This is a subtle shift from the typical approach of changing a few words of the source text to paraphrase it. This is also great practice for honing your sentence and paragraph writing skills.
I have built a proof of concept that demonstrates the core mechanics of the approach. The challenge is that my background is in backend development, and so the UX doesn't make a good first impression. I am building a more polished demo, but it is slow going given my lack of frontend expertise.
I can so relate :). I left my position as a principal engineer at a FAANG/MAAMA/etc. company in 2020 with a general direction to try to make a difference in the education space. It has been maddening trying to live up to my own expectations.
>> what's the best way to statistically guage a student's knowledge level?
As you have probably already discovered, there is quite a bit of research in this area :). If anyone else is interested, you can search Google Scholar for "deep knowledge tracing", "bayesian knowledge tracing", "performance factor analysis", "knowledge space theory", "intelligent tutoring systems", and "item response theory". Some intelligent tutoring systems include ALEKS, Squirrel Ai, and Riiid.
According to the National Assessment of Educational Progress, only 35% of U.S. fourth graders in 2019 could read proficiently [0]. The pandemic has only made this worse [1] [2]. Once students start to fall behind in reading, they fall further and further behind their peers each year. What is, perhaps, most criminal is that there is a mountain of research [3][4][5][6][7] about what works, and yet these approaches aren’t employed nor even well-known.
I am a former principal engineer at a FAANG/MAMAA/etc. [8] company turned independent educational researcher, and I believe I have developed a novel edtech approach to address this learning gap. I am collaborating with a few data scientists and educators, but am searching for UX designers, curriculum designers, writers, and marketers who can help make this a reality.
In brief, one of the core challenges in edtech is how to build an engaging app that actually results in learning (many apps resort to psychological tricks [9] and gamification hacks [10] that are long-term detrimental.) Although I have a high-level direction for how to achieve this, my skill set is mostly in backend development and public speaking.
If you are interested, please reach out to (my HN username) @ (Microsoft's not cold email domain acquired in 1997).
Is there a whitepaper that documents how PrepScholar "automatically learns the strengths and weaknesses of each student and creates an individualized learning program through machine learning"? For example, ALEKS [0] is based on the Knowledge Space Theory [1].
> I think you have an up hill battle with getting traction with DI, but if you succeed that would be a great benefit to many people. (my perception)
I completely agree and it really isn't too surprising that DI never took off in schools. FWIW, I am looking to revive the underlying pedagogy (e.g., teaching sequences for non-comparatives, joining forms, etc.) rather than DI proper.
> What is the protocol on this site for starting conversations "offline"?
Interesting question that I never stopped to consider :). I have met dozens of people through HN, and it seems congruent with the hacker ethic ("throughout writings about hackers and their work processes, a common value of community and collaboration is present"), but ::shrugs::.
Thanks again for sharing your DI story.