Ramblings on language learning

Learning is a complex word and its meaning is becoming increasingly vague, nowadays even more with all the AI Hype going on, can machines learn? can only biological beings learn? what exactly is to learn?

Those are hard questions and when it comes to language “learning/speaking” I am afraid there is no straight answer neither, and the word can mean many things: from a youtuber trying to get views bragging while saying a few sentences, to a more concrete attempt like the Common European Framework of Reference (CEFR)

This makes it difficult to start from a common reference point, but if you ask me, the answer lies more towards the independent and spontaneous use of the language which would be something from a B2/C1 level onwards on the CEFR reference framework. So for the purpose of this experiment, the general idea is to test how much can you learn/speak after a 3 month test.

I grabbed some ideas to try from this Tim Ferris blog’s post, which I highly recommend , but the general inspiration came after being constantly exposed on my last vacations to Dutch and also just for fun!

I already speak ( or use on a daily basis?) german and english so dutch sounds like something you were supposed to understand but you don’t, and If you have heard it before you would notice those weird noises they do with the throat like kind of a person is having a mini choking/ clearing their throats. So for me the starting point would be to figure out how to pronounce that and how to recognize what is being said when I hear it or worse see it written.

This made the perfect occasion to try language learning apps; for me they are pretty new, and I have been more a pen, paper and that audio recorder every teacher uses guy, so yes I know the pain.

I did a small search on the app store and I didn’t really think much about it and went with Duolingo I had seen before some people using it, as well as the owl memes, and it finally came under my radar when they announced the new GPT-4 integration. You start translating some basic phrases and they give introduce you more words progressively. They also gamify it so there are some challenges, leaderboards and games which makes the app catchy.

However, as I used it daily even though I liked it, and it hugely helped me getting started and with the basic sounds, I had this feeling it was going too slow; some exercises started to get repetitive and sometimes kind of restrictive, probably I also wanted ti get rid of some weird phrases too like I am eating bread and crying on the floor (too many tears already)

Exploring the frequency lists

So I went with Tim’s idea and got a list of the most frequent words in dutch. The basic idea is that there is a point of learning you can reach while every portion of time allocated to learning translates directly in progress, and after that point you need to spend much more time learning for just a small progress (law of diminishing retuns)

To understand 95% of a language and become conversational fluent may require 3 months of applied learning; to reach the 98% threshold could require 10 years

So the idea is to look for the best return per hour invested- ROI and for that we search for the most used/frequent words in any language. For the case of English:

The first 25 most common writen words make up about 1/3 of all printed material . The first 100 comprise 1/2 of all written material, and the first 300 make up about 65% percent of all written material in English

So that’s great ! there are some words that you commonly use and that make up a vast part of the vocabulary you would normally use, and If you think about it, it makes sense. We all use the word the, eat, I far more than exarcerbate or ultracrepidarian.

Given this fact, I used a frequency dictionary in this case for dutch and searched for the 72 most common verbs, then for the 72 most common adverbs, then adjectives and finally (conjunctions/ prepositions), I decided to make flashcards with them and to study them everyday using a spaced repetition software (Anki )

Creation of the Deck

I created the decks as simple as possible with the dutch question and the english translation on the back for those 72 items of each mentioned above, but I also made one deck for general conjugation and another with present, past and perfect tense for those 72 verbs.

The idea was to spend from 15 to 30 minutes a day reviewing the cards but I also run some simulations with an extension, to check how much of the deck would I could cover, according to different settings like more cards in a day, see the easiest one less frequently and so on, the idea was just to experiment with it.

Grammar and Six Lines of Gold

According to Tim, you should deconstruct the language before starting. This is to consider them like a sport. There are some prerequisites, some rules and according to that you can have an idea of how long it will take for you. Here are six questions he used from the outset to start the deconstruction:

1. Are there new grammatical structures that will postpone fluency? (look at SOV vs. SVO, as well as noun cases)

2. Are there new sounds that will double or quadruple time to fluency? (especially vowels)

3. How similar is it to languages I already understand? What will help and what will interfere? (Will acquisition erase a previous language? Can I borrow structures without fatal interference like Portuguese after Spanish?)

4. All of which answer: How difficult will it be, and how long would it take to become functionally fluent?

You can make the exercise according to your target and needs, try to reduce it to the most essential features and to mark your goals as precise as possible, this can be easier said than done specially if you are just getting started it may be not that intuitive what you will need or how to get there, but the exersice itself may be helpful to keep you on track.

Reading, reading, reading

There are many ways to apply what you have learned, for one thing is the passive learning act like going t o class, listening casually to something, seeing a word on the board, but the active learning part is where you can test how much you actually know and where are your gaps. Things like writing a text, having a conversation or reading a text either reinforce your learning or give you new insights.

Some people have the possibility to move to a new place where that language is spoken, making it easier, to adapt it to your routine, some others know a person(s) with whom they can practice on a frequent basis. Nonetheless this is not always the case, and one of the strategies I think are also useful regardless of your situation is reading in your target language. In The inescapable case for extensive reading the authors argue how graded reading is an “indispensable part of any language program” and they also explain it in a statistical context like with the image given bellow:

what they are trying to measure here is how many words need to be read to learn all the 500, 1000 and so on most frequent words in English. The basic idea is: you can build a vast vocabulary simply by reading.

So for this to be more effective you have to aim for books and texts aimed to your level and keep aiming for new and longer texts as you progress.

In my case I started with child books like the Aesop’s fables after my basic vocabulary review. Kid books offer pretty good material to start and there are also books graded by level like A- Beginners, B-Intermediate ones where they adapt the texts specifically for this and you can even find adapted versions of literature classics.

Summing Up

There are many learning strategies and everything depends on your personal situation, as a general rule you should be aware of over-promising and scams, nevertheless there are indeed some ways to make your learning process more efficient and some of them were presented on this article. Many of the things mentioned here deserved their own article but for the sake of brevity I tried to keep it short, I hope that anything mentioned here may inspire you or help you with in something. Also general experimentation is a good thing and you may try it too, May the Force be with you on your learning journey.

Hej, it’s been a while since I am in a toxic-dependant relationship with the german. After each step it gets harder and harder to reach the next one, because the time you need to get a small improvement increase drastically. It feels like after running a marathon to the exhaustion,you realize you are only half way through.

Once you get to the C level (Kompetente Sprachverwendung) you basically have two choices: Leave it as it is and try to maintain it, maybe losing it slowly or you could go the extra mile and try to master it (Annährend muttersprachliche Kenntnisse). So even though its a hard nut to crack you can still get away with it, and hopefully with the least number of traumas and life crisis.

In this article I will describe a few learning strategies for the amelioration of german specially aimed for more advanced levels using spaced repetition and active recall.To start I checked all my notes and some copies I still had from previous courses. Taking a look I see that there are some small details that I might had forgotten or overlooked, also I needed to finish memorizing some key structures.

Making the list, I saw that what I wanted to review was this:

• The list of Starke und unregelmässige Verben
• Funktionsverbgefügen
• Verben mit direktem Kasus (Nom/Akk/Dativ)
• Verben mit präpositionalem Kasus
• Wissenschafts-und fachsprachliches Training
• Verben mit un/persönlichem Subjekt + Objekt

As a method, I decided that the notes I had, beyond provinding some good memories from the time I was learning would not be of much use. It is normal to forget stuff and we are used to being stuffed with information to later vomit it up in the exams without really digesting it. Usually they never explain you how the brain and memory works, so I had to ask myself: what was I doing wrong? why? Its not a secret to anyone not only in german but in life itself, that spending many hours doing something not always translate into an improvement, o sometimes we get overwhelmed with the stuff and after trying it a couple of times we end giving it up.

Therefore in this case after reading about the forgetting curve and Leitner system (for more information on this:) https://ncase.me/remember/; I decided to apply it for this case, due to the fact that it works perfectly for informations that you want to retain in a really long term. The system can be done manually or using a software, and there are several available using different learning algorithms. In my case I chose Anki due to the fact that is open source and available in every platform.

Deck Building

As I was just starting, I didn’t want to focus on the technical details or doing it perfectly, instead I just focused on starting and moving forward correcting details in the way. The goal with Anki is to create question cards and at the same time review them. However there are several card types and theories about what should a card contain. For example in a verb list you may want to include all its forms for a given verb in a card or you may want to split it in several cards for each verb form.

I also wanted to use colors as a reference and to each gramatical case I assigned a color to help my myself to identify them.

Blue is key and will be used a lot, however a small trick here: due to the fact that yellow(Akkusativ) appears much more than the blue(Dativ) and green(Genitiv), learning the blue and green first will be helpful to guess the yellow without learning all of them at once.

Starke und unregelmässige Verben

In this case we would get 2 types of cards: one with the verb 2 language- german and another with the three grammatical of the verb

Verben mit direktem Kasus ( Nom/Akk/Dativ)

In this case I took the paper list I had, and transcibed it into a spreadsheet, and as mentioned before classifying it by colors giving priority to the blue and green. However in the list there is also the combination of double yellow (Verben mit zwei Akkusativen) yellow-green (Verben mit Akkusativ and Genitiv) and yellow-blue ( Akkusativ und Dativ)

Funktionsverbgefügen

Please don’t run away! this is the hardest part and to be honest the one that really makes you to ace it out, those are little details that make the difference. After spending a long time searching for them and putting them on a spreadsheet, I got a list of approximatly 273 nominal verbs, which I considered worthwhile knowing. The idea was also to find their meaning and an example in german that could help me to get in context and lastly a translation to better understand it, for example:

In this case two types of them are used (Typ 1 /Typ 2), where in the first one the substantive and the verb which it represents have a similar meaning, and in the second one the verb and the substantive represented differ more in meaning.

Verben mit Präpositionelem Kasus (Präposition + Akkusativ / Präposition + Dativ)

This list is really similar to the list of direktem Kasus, only that here a preposition is added with the case. Many time by knowing the preposition we could foresee which case will it use, but many times we find ourselves with surprises or sometime a verb can use several prepositions and change completely its meaning, so even though you might have seen the verb before when you find yourself with one of these you could get lost ( sometimes the context may save you)

Verben mit un/persönlichem Subjekt + Objekt

Here we want to review the personal and impersonal verbs and the case which they use

Wissenschafts-und fachsprachliches Training

In this case the cards where aimed more to phrases used in college, exams and advance level books. Usually those are list used in presentations and academic writing.

In total, I got more than two thousand cards and the idea is to space the reviews, depending on your goals you may configure the number of daily cards, the spacing and many things, you may even simulate your progress and check if it adjust to your goals

Active recall, spaced repetion, mind mapping and interleaved practice are tools that may help you with your learning regardless of the subject. Here I wanted to give you a quick glimpse in how I use them in a later stage of language learning, but there are many more uses to them and a good complement to your learning.

“ (GDP) … measures everything except that which is worthwhile”

“Yet the gross national product does not allow for the health of our children, the quality of their education or the joy of their play. It does not include the beauty of our poetry or the strength of our marriages, the intelligence of our public debate or the integrity of our public officials.”

Nowadays the debate is still open. Does it really show how good or bad the economy is performing? Is it measured accurately? Can it really comunicate us something?

I am afraid there is no simple answer for that question, and in this article I am going to try to ask the genie of the lamp if there is any relationship between the gdp and the life expectancy. The dataset was obtained from Codecademy and it contains the gdp and life expetancy for 6 countries for a period of 15 years .

You may find the complete code along with the csv and jupyter file in this repository: https://github.com/eriglesias/GDP_Life_Expectancy

Goals

• Has life expectancy increased over time in the six nations?
• Has GDP increased over time in the six nations?
• Is there a correlation between GDP and life expectancy of a country?
• What is the average life expectancy in these nations?

Data Exploration

To start we loaded the data from the csv file and take a glimpse of it, we wanted to know which countries and what period of years were involved. In this case we have 6 different countries for this data set: 3 located in America 1 in Europe 1 in Africa and 1 in Asia and the years goes from 2000 to 2011.

Also to get a better grasp of the data we do some exploratory plots just to undestand better its distribution, skewness, modality and check for outliers

Analysis

As we had the data for several countries and several years the best way to work with the data was to group it, doing so we got:

doing this we created the average values for GDP and LEB grouped by country for an initial anaylis, which could be seen in the following plots:

Doing this we can take see, how the live expectancy in the world for most of the cases is above the 70’s except for Zimbabwe which is falling almost 20 years behind. Also we can see how massive is the GDP of the United States in comparison with the other countries in the data set (which doesn’t account for a equally massive difference of life expetancy with the other countries). As the GDP for Zimbabwe is so small it can not be seen in the plot. Now we want to see how over the years this two measures have grown.

Taking the wholeness of the data we can see how the countries with the bigger gdp growth were China and the U.S, however the country with the most notorious GDP growth was Zimbabwe. However the country with the most notorious life expectancy change was Zimbabwe.

To see if there is a correlation between GDP and Life expectancy we plot the both:

We can see here how the GDP is correlated with the life expectancy, many of the countries present a linear relationship however China present more an exponencial one. In the case of Zimbabwe the life expectancy shrinked a while, but after the 0.6 trillion barrier it presented a really accelerated growth shortening the distance with the rest of the countries.

Conclusion

Data presents a lot of opportunities to make better conclusions, discover unseen patterns, ask better questions and much more. Data Science and Data Analysis is a growing and interesting field to learn or be considered.

In this article I wanted to present a small glimpse on a data analysis data set, using exploratory data anaylisis, finding a question to solve, gattering the data, looking what kind of data we have and looking for outliers or dataset changes. The libraries used to do the visualization were Matplotlib and Seaborn, you may find the complete version of the code and anaylisis in the following repository: https://github.com/eriglesias/GDP_Life_Expectancy