Using Learning With Texts to compare language complexity / irregularity

I wrote yesterday that I've just installed Learning With Texts and have been comparing it with Lingq. It's actually quite different in feel, and I don't think I'd recommend one entirely over the other. LWT does allow for a lot more experimentation though. After adding German I thought I would check out something I've been wondering for a while: how much help would it be to entirely memorize a text? Since I know every word in Demian (which I translated and proofread about a dozen times) I thought I'd start with that. This is possible because LWT allows you to check a button saying you know every word in a text. So here are the results when I entered Demian one chapter at a time, checking the I know every word button each time.

First chapter: 1806 unknown words.
Next: 919
Next: 917
Next: 1005
Next: 726
Next: 677
Next: 791
Next: 239

Then the book ended, and so did the experiment. You can see though that the number of new words dropped right away in the beginning, stabilized a bit, and then eventually dropped again. Not sure what would have happened if the book had been longer.

After that I decided to use it to compare the complexity / irregularity of two Germanic languages, German and Afrikaans. For that we'll need a biblical text, and of course that means Ecclesiastes - the best book in the Bible and fairly short at 12 chapters. By comparing complexity or irregularity I mean this: how many new words, or different forms of words, will the student have to learn to understand the text? You may know the word fight but unless you know that the past tense is fought and not fighted, you still have something left to learn. Same thing with German: you may know schlafen but if you see the word schläft and don't know that it's the same word, you haven't entirely learned it. Afrikaans with almost completely regular verb conjugation, no cases (except in pronouns) and no grammatical gender would probably require less from the student compared with German. So I ran a quick test to find out.

The numbers below represent the following: number of new words to learn, with total/ number of words per chapter. One line below is the percentage of unknown words as our student reads through the book, learning each and every word perfectly.

Chapter 1 - German 185 (185) Afrikaans 142 (142)
German 100% unknown, Afrikaans 100% unknown

Chapter 2 - German 198 (284) Afrikaans 160 (234)
German 70% unknown, Afrikaans 68% unknown

Chapter 3 - German 99 (211) Afrikaans 76 (177)
German 47% unknown, Afrikaans 43% unknown

Chapter 4 - German 97 (194) Afrikaans 68 (175)
German 50% unknown, Afrikaans 39% unknown

Chapter 5 - German 105 (255) Afrikaans 73 (188)
German 41% unknown, Afrikaans 39% unknown

Chapter 6 - German 50 (159) Afrikaans 31 (126)
German 31% unknown, Afrikaans 25% unknown

Chapter 7 - German 123 (281) Afrikaans 93 (243)
German 44% unknown, Afrikaans 38% unknown

Chapter 8 - German 73 (221) Afrikaans 55 (177)
German 33% unknown, Afrikaans 31% unknown

Chapter 9 - German 90 (260) Afrikaans 61 (203)
German 35% unknown, Afrikaans 30% unknown

Chapter 10 - German 91 (224) Afrikaans 63 (183)
German 41% unknown, Afrikaans 34% unknown

Chapter 11 - German 46 (156) Afrikaans 35 (131)
German 29% unknown, Afrikaans 27% unknown

Chapter 12 - German 87 (202) Afrikaans 69 (178)
German 43% unknown, Afrikaans 39% unknown

You'll notice that every time the student of Afrikaans has to learn both fewer words in total as well as fewer words per capita than the German student. I plan to run a more extensive test like this on the weekend (maybe with five chapters at a time, or using all the Ted.com talks in Afrikaans vs. German), but I think the results show what should be common sense: there are fewer things to learn in Afrikaans compared to German.

A graph showing the unknown percentage of words as the student progresses chapter by chapter:

More on this in a few days!