The way one uses language at home is extremely narrow compared to the range of ways one may use language in the outside world. Hence kids who grow up bilingual but don't use their parents' language outside tend to have a very limited command of it
Speaking a language is not a binary, it's a multidimensional spectrum. Not only that, but speaking, understanding speech, reading, and writing are all distinct skills, and while they're highly synergistic, they overlap a lot less than one may think
One may be able to read without being able to speak, or even without being able to understand speech. And inversely. It's also a lot easier to passively understand than to express oneself
Another interesting distinction is the one between freeform writing (pen & paper) and selection-based writing (using a keyboard). It takes one order of magnitude more time and memory capacity to learn to write Japanese or Chinese on paper that to type the exact same characters
In the future it's likely very few people will still be able to freeform-write these languages. Even if they keep painstakingly teaching these skills in school, people will forget them quickly enough afterwards due to a lack of usage

• • •

Missing some Tweet in this thread? You can try to force a refresh
 

Keep Current with François Chollet

François Chollet Profile picture

Stay in touch and get notified when new unrolls are available from this author!

Read all threads

This Thread may be Removed Anytime!

PDF

Twitter may remove this content at anytime! Save it as PDF for later use!

Try unrolling a thread yourself!

how to unroll video
  1. Follow @ThreadReaderApp to mention us!

  2. From a Twitter thread mention us with a keyword "unroll"
@threadreaderapp unroll

Practice here first or read more on our help page!

More from @fchollet

14 Dec
Humans develop their full cognitive potential in an environment that is complex & challenging, without being overwhelming. Similarly, the big technological leaps of past civilizations have occurred in response to environmental constraints that were challenging, but not too harsh.
A lack of challenges and hardships is just as big an obstacle to the realization of one's potential as facing hardships so tough they cannot be overcome. This applies to individuals and cultures alike.
In the first few millennia of the history of civilization, natural environmental constraints were the main driver of (and limit to) human ingenuity. New technology arose from the need to survive in challenging environments.
Read 6 tweets
6 Dec
Deep learning for auto-captioning 👍👍👍
This said "abstraction and reasoning in AI systems", but the alternative version is actually more interesting
The *Ice systems* is the name of an extensive network of galleries on the icy moon Europa, built eons ago by a long-gone culture. In the cavernous tunnels, a mere whisper can echo over incredible distances...
Read 4 tweets
2 Dec
~277k dead from Covid in the US so far. Real number probably somewhat higher. Will be over 500k by the time it's over. It didn't have to turn out this way
This is obviously the biggest news story this year. The second biggest being the continued march of climate change and the devastation in its wake
The attacks on democracy and the rule of law only make it to third place, so it's kind of a bad year
Read 4 tweets
21 Nov
One of the most difficult things to master in a foreign language: cases where the target language uses multiple words (subcategories) for a concept for which your previous languages only have one word. You're simply not used to parse reality into these new subcategories.
A simple example would be counters in Japanese. In European languages, a counter is just a number, so you would say, "three birds, three peanuts, three sheets of paper, three pens, three cars". In Japanese, a counter conveys not just quantity but also various object properties
So you would say, sanba (birds and bunnies), sanko (generic small thing), sanmai (flat thing), sanbon (cylindrical thing), sandai (large vehicle), etc. In fact there are more distinct counters than I can count. It takes a long time to get used to these distinctions
Read 5 tweets
20 Nov
A few neat things in the next TF release:

- The Keras mixed precision API moves to stable
- Multi-worker mirrored distribution support moves to stable
- Experimental support for parameter server distribution in Keras
- Full NumPy API implementation (tf.experimental.numpy)
Using mixed precision: tensorflow.org/guide/mixed_pr…
Using parameter server distribution: tensorflow.org/tutorials/dist…
Read 4 tweets
15 Nov
The territory, the map, and the cartographer each belong in entirely different categories. Yet, we tend to confuse them with each other

The thing, the abstraction that captures an actionable aspect of the thing, and the intelligence that is capable of generating new abstractions
In particular, don't confuse a model and the ability to produce new models. Humans excel at operationalizing the abstractions they invent into highly effective machines, but so far our machines have close to zero ability to generate abstractions of their own
A road takes you from A to B, but a road-building company takes you from anywhere to anywhere else.

An operationalized abstraction solves the task for which it was designed, but intelligence -- the ability to produce abstractions -- can solve arbitrary tasks
Read 4 tweets

Did Thread Reader help you today?

Support us! We are indie developers!


This site is made by just two indie developers on a laptop doing marketing, support and development! Read more about the story.

Become a Premium Member ($3/month or $30/year) and get exclusive features!

Become Premium

Too expensive? Make a small donation by buying us coffee ($5) or help with server cost ($10)

Donate via Paypal Become our Patreon

Thank you for your support!

Follow Us on Twitter!