645 views 0 comments

The best way to learn a new language

by on October 10, 2013


We brought in a fellow linguist at the lab for this article.

We used to teach languages almost entirely through translation, vocabulary, and grammar exercises. This made sense when the languages were Latin and ancient Greek, because you don’t need to learn to communicate in dead languages. Schools in Japan are still largely stuck on this model, although they’re getting a lot better.

The Behaviorist revolution brought about a huge shift toward “habit formation”. Under the theoretical framework that all behaviors are essentially habits, language learning was seen as the process of abandoning habits from the L1 (native language) and developing habits from the L2. The Audiolingual Method is a good example of this era. Students drill, drill, drill, and get quick and constant corrections from the teacher. Conversation skills didn’t matter, or rather, they were supposed to come naturally from the development of correct habits..

Chomsky developed a Generative theory of language, meaning that language is created by each individual, rather than learned from external sources. The idea is that we all have an inherent capacity for language, and certain aspects of language (nouns, adjectives, recursion) are inherent to all languages and “hard coded” in the human brain. In this view, habit building is bullshit, because the student needs to be guided in creating his own language. This is still the dominant view today.

In the mid-80s, Krashen developed a model of SLA called the “Monitor Model“. It’s a bit complex if you don’t have a background in SLA, but his basic advice to teachers is fairly simple: Studying grammar is pointless, because you’ll only learn how to talk about the language, rather than use the language. Instead, students need to get lots of “comprehensible input”, or language that they can understand, that is just slightly above their current level. This is still a very influential strain of thought, but it doesn’t have a lot of experimental evidence to back it up.

Language Learning Today
These days, we consider SLA to be in the “post-method” era. Most people don’t think that there’s any one, perfect way to learn a language. However, most people agree that you need:

  1. Input – Learners need to hear spoken language and read written language. It needs to be comprehensible, challenging, contextualized, communicative, meaningful, and relevant to the student. Duolingo’s input is okay, but it’s not contextualized, relevant or communicative, and often not meaningful. “El pajaro bebe la leche”. What bird? Why? Is there a new type of bird that’s turning into a mammal? Why are you telling me this?
  2. Output – Learners need to practice speaking and writing to develop those skills. Learners aren’t pushed to talk about things that are a little difficult for them don’t seem to acquire fluent, grammatical speech (see Swain’s studies on Canadian immersion programs). Duolingo has a little bit of writing, but it’s all just translation. There’s no writing to express your opinion, or ask for a favor, or anything like that.
  3. Communicative practice – It’s not enough to drill patterns; the language has to be used to communicate with other people. This is my primary beef with Duolingo.
  4. Some amount of direct study – Kids can pick up language from their surroundings, but that doesn’t seem to work for adults. Adults need to use their higher cognitive skills to engage and analyze the language. This is basically what Duolingo does. However, this probably accounts for a very small part of language learning–in my own acquisition of Japanese, I think I’ve spent less than 1% of my time analyzing sentences or translating.

How I Learn and Teach
As an applicable approach, I like the Skill Acquisition Theory. Basically, you start with knowing about something, then learn how to use it, then practice until it becomes automatic.

This is more or less how you teach someone to play a musical instrument. Think about someone learning to play guitar. Of course, you have to learn that the strings are EADGBE, and you have to learn that putting your finger on the first fret makes it go up a half step, and so on. You have to learn what a scale is, and what an arpeggio is, and all that. You also have to learn the physical skills, like strumming, hammer-ons, and wicked sweet finger tapping. And then you have to practice for around 10,000 hours if you want to kick ass at guitar. In terms of time spent, it’s something like 2%/5%/93%. I think using roughly similar ratios makes for a successful language learning experience.

There is more to language learning than grammar and vocab. It’s a communicative skill, and you won’t be very successful if you try to learn it without communicating or by treating it as an academic subject.