There are 2 main differences. One is found in methodology, the second in psychology.
This image represents how a student typically learns a language when they go to a language class. As a former language teacher and language student I'm familiar with the drill of many types of language practice.
Normally, language practice is structured on teaching the forms of grammar considered the least complicated and the most usable in conversation. Almost immediately, people will learn to talk about the present, for example. Describing a past day follows almost shortly after.
Although in principle the idea I think is acceptable for some people, I find some problems with the nature of this style of methodology:
- Vocabulary is usually chosen in an artificial manner about what the textbook writer (or teacher if you're lucky) thinks is needed. Secondly most language teachers only focus on teaching "correct" or "proper" language rather than what is most frequently encountered. Thus I had French students who despite having learned English for over 40 years, didn't know things like "gonna" and "wanna".
- Focus on immediate production without an incubation period. Production without first learning the norms of the language gives the student no reference to work on. It is like getting a person to write a paper on macroeconomic theory having never studied economics.
While most language methods agree you need exposure to all elements in the above circle, many argue about which you do first.
This argument I believe comes down to psychology and should be treated as such. It is not in the domain of linguists.
Linguists study language structure, origins, and the interplay between language, culture and thought. But they do not study the how humans learn.
One of the first functions we develop as babies is an ability to find patterns. Through observation we learn to infer information from these patterns. It is through this ability that children learn their master language(s).
Later on at about 10 years of age, our cognitive skills develop which enable abstract thinking. The basis of the critical period hypothesis is that this cognitive dominance inhibits the ability to infer, and hence learn languages the way children do.
If this thinking sounds flawed to you. I totally agree. There is a serious lack of scientific process applied to this research. In the scientific method, to identify the influence of a single factor (like chronological age), all other variables should be kept the same if possible. In telecommunications, we used to refer to this as isolation testing.
To support the critical period hypothesis, a quality scientific study would need to use the same methodology children learn languages with adults. I have not found one study yet that does this. I've looked.
Common sense dictates that our ability to match patterns doesn't diminish either. Not if we live with our eyes open. Yet thinking persists in language (amongst other subjects) that we need to have a rule and reason for each lesson. In reality it just satisfies our brains need for a reason. It doesn't help much with gaining ability in a language though.
This picture represents language acquisition methods. A key difference is the order of the approach. Acquisition, natural, or communicative approaches focus first and foremost on providing accompanying context and environment that is conducive to inference, or "guessing" of meaning.
It is an implicit method, rather than the explicit method above that gives certain sounding rules. These rules are often comforting to adults who have a fear of uncertainty, but they are limiting as rules often only function in a restricted fashion.
Acquisition methods instead focus on the acquisition of norms as opposed to rules. Norms are rooted in the social structure of a group. In language learning, the native speakers form the 'group'. Like becoming a team member in a workplace, we need to learn the norms of the group to be in the group. Otherwise we will always be an outsider.
Acquisition methods like ALG seek to create an environment that maximizes comprehensible input and allows the student to infer meaning through observation of patterns over time. This requires the student to attempt to not analyze the language and as a result they are not allowed to take notes in class.
There is also a silent period which allows the student to identify the norms of the language before production begins. If the student hears a lot of non-native usage of the language, they become unable to differentiate what sounds normal in the target language, and what does not.
The method itself requires a leap of faith because it flies in the face of what most adults believe is "learning". In fact the process engages a very different area in our brain.
This is how people who are uneducated and often not cognitively developed (such as poverty stricken refugees from Myanmar here in Thailand) can become fluent in a new language even if they move as an adult. This doesn't happen with all, but it does with those that interact frequently with the local population (such as Burmese in service roles). For this also there is insufficient research conducted about how these groups learn languages.
Acquisition methods also may be a better way to learn vocabulary. Meredith Brinster of John Hopkins University has recently conducted research suggesting that children retain vocabulary better through inference than through explicit teaching. I'd like to see research like this done on adults.
I'm not saying that language learning doesn't work, but that it is getting the order backwards and using the wrong neural process to learn in an ideal fashion. It focuses on output before significant amounts of input are gained and not exposing the students to linguistic norms until they are advanced students. This results in the student transferring the linguistic norms of their own language onto the second one, resulting in a student that will always speak like a foreigner.
As David Long of AUA Thailand says, "ALG is a foundation method. There is no other method that takes a person up to a level where production should occur naturally".
If you are a language learner, I strongly recommend maximizing your amount of comprehensible input as I am trying to do with my experiment on acquiring Chinese.