That is, they either strongly believe that a learner should maximize their output, or maximize their input in order to best start learning a language.
I fall into the input camp. I believe it's important to maximize input as much as possible in the first 6 months to 1 year before the learner starts to focus on output. Without having prior experience with the language, then when it comes to production the learner has no reference to draw on to form utterances.
Often, rules just don't cut it.
I present my arguments for each of the four main components of language.
Pragmatics
This is the process of connecting language with context. In such a way something like "whatever" can have different meanings depending on the context of which it is used (such as declaring your indifference, or indicating that you want a person to shutup).
This is almost impossible to learn except by exposure to real life situations, or through TV shows and other video input. Most supporters of output place this kind of learning as a final stage. I suggest starting exposure to this kind of input from day one. Many complex sentences will not be understood at first, but simple ones can be picked up fairly quickly. For example, in 50 hours of Chinese exposure I've learned 2 ways of saying "let's go" depending on urgency and the verb "to go" which is different.
I feel that with enough exposure, this creates a link with certain phonetic patterns and situations (like identifying levels of formality by phonetics, describing an academic subject compared to describing a sport, etc).
Phonetics
The output argument generally tends towards a listen-and-practice approach. The idea being that phonetic sounds need to be practiced over and over again until the vocal chords and mouth learn to make the sounds required in the new language.
What inputters believe, and as my experience has shown, having a good level of input creates an instinct for the correct sounds of the language they are learning. It also gives them time to tune their ear to the phonetics, refining their ability to hear nuances in the sounds. This allows them to self correct in a far more efficient way than a listen-once-and-repeat method does.
Not only that, but often sadly most non-input methods tend to study phonetics in isolation, and not often how they combine together to make sentences.
The phonetics of a language involve far more than the individual consonant and vowel sounds. There is also syllabic stress (think of "REcord" and "to reCORD"), word liasons ("it's not" or "it snot"?), dropped vowels/syllables ("watchya doin?"), suprasegmental stress ("did YOU drink the milk?" or "did you DRINK the milk?"), emotional tone and formal tone, among other things.
These are all extremely involved and would be rather complex if studied in separate segments while trying to remember a rule for all of them. But they are present in natural speech, all happening at once. As such, the same input can be used many different ways to study the language from a different perspective.
Grammar
From a modern language learning perspective, this is considered the foundation of learning. It is usually the first thing you learn as it can be used to create comprehensible sentences straight away.
The problem with this is that often this can be so complex that it takes the main focus of the student away from comprehension of the language and the student starts to think of language in a mechanical fashion. Some grammar can be simple and can be practiced enough that it can become automatic. But there are some grammar forms that are context specific or difficult to explain in an explicit format. Sometimes they are classed as expressions, but the line is not always so clear. For example, "What's up with Matt?" may be considered an expression but "What is the problem with Matt?" may be considered a standard form of grammar, even though they are both interrogatives.
In addition, grammar cannot be separated from context. In the example above, "What's up?" is an informal expression inappropriate for some uses. If you learn this in a textbook it might not be clear. But with input, when this is viewed as only being used among friends or in casual situations, the person can intuit that this is an informal expression.
While learning grammar by rules can be useful, I believe it is best used to refine knowledge already present, like a native speaking child does when they go to primary school and beyond.
But even amongst native speakers, the strongest determinant of good spelling and grammar usage is how much the person reads. In other words, input.
Vocabulary
Lastly vocabulary is best learned by input when possible. A person can learn vocabulary through studying lists, through flashcards, picture cards, or through natural input.
Through viewing conversations in real life or even from TV or the like, the person has more senses engaged in which to remember vocabulary. Some people are visually focused, some auditory and some tactile. By giving the person something to look at, hear, and touch if possible it allows them to use the input method that is strongest for them. And by providing the context of the situation and supporting conversation it adds additional information (like when to use "ass" or "butt"; or "boat" or "ship").
--------------------------------------------------------------------------------------------
What is the weak point of the input argument?
For language acquisition purposes, the language needs to be as comprehensible as possible, especially at early stages. As the guy that taught me TESOL said, "don't explain, demonstrate". It is not always very easy to find input like this, and it can be hard to convince native speakers to teach like this (with methods like TPR for example).
The debate is multifaceted and there are many justifications on both sides. Ultimately though, but input and output need to be thoroughly practiced. As a language learner what you need to make sure is that in the long term you are balancing input and output. Then focus first on what seems to work best for you.
