Looking for better methods for learning languages I decided to dive into How Languages are Learned by Patsy M. Lightbown & Nina Spada.
The book is dedicated to teachers, and as such, offers superfluous information for a person simply interested in learning. Nevertheless, the book is simply packed with information that is also relevant to a student.
The text is divided into 7 chapters and follows a very natural pattern: introducing first language acquisition in the first chapter; the problem of second language (L2) acquisition in the second; individual differences in L2 acquisition; the main psychological and cognitive theories second languages acquisition; how classroom-based studies are performed; and various studies about L2 teaching methodologies. The book concludes with a chapter summarizing the various popular (mis-)conceptions related to language learning.
The authors cite studies for each of their claims, and each chapter ends with suggestions for further reading. I really appreciate the amount of work that must have been put when writing the book, yet despite the scientific rigor the book doesn't have a scholar as its target. Concepts are introduced gradually and there's no problem in following the matter of the book, even if one doesn't have a background in linguistics or theory of education.
One of the theories covered in the book is that language learners develop an intralangue as they progress, their working hypothesis on how a language works. Sometimes, learners may produce more errors then before simply because they unify some rules and generalize previous cases that were learned in whole (chunking) e.g "He went" may change to "He goed" as a student learns that the past form of verb is formed by adding the +ed suffix, instead of pronouncing "He went" as a formula.
The authors propose three stages of language skill - declarative knowledge, procedural knowledge and automatic knowledge. Declarative knowledge are the rules about the language a student might posses but doesn't know how to execute in contrast to procedural knowledge. Automatic knowledge is procedural knowledge that is refined to a point that it stops being a conscious effort.
Students use different means of acquiring language. Some exceptional cases can exhibit even near native-level performance no matter the learning methodology but for most cases near native-level performance is only achievable if learning had begun before the age of 15. Diverse teaching methods combining both structure based (grammar rules) and communication based (comprehension and producing speech and writing) methods can also facilitate learning, wheres concentrating only one aspect of the language (comprehensions, speaking etc.) gives mixed results.
Most adult students never achieve near-native pronunciation, although there are exceptions, and according to studies stress and rhythm are more important for understanding than pronunciation itself.
Passive sources of language such as books, television or radio seem to not be sufficient for learning a language, even for native speakers. Though, reading for pleasure seems to be the best way of increasing one's vocabulary. Procuring books adequate for a student might be a difficult task - for a text to be comprehensible at least 90-95% of the words need to be known. And according to studies, a new word needs to be seen at least 16 times to be remembered, although activates that require them to attend carefully to new words and use them in production tasks might speed up the process of learning.
I would recommend How Languages are Learned to anyone interested in the topic of language acquisition. The text is very approachable (though targeted to teachers), and even though some of the information given in the book might not be instantaneously applicable, the amount of useful knowledge is just incredible!