Ideally the model and its inductive steps should start soon after birth, with the learning of motor control. The starting point should be the capabilities that we are comfortable with claiming that they are encoded in the genes and thus part of the inherited structure (the engineering design) rather than being acquired or learned.

Alternatively one can analyze language as representing a symbolic encoding scheme.  Words, sentences, and paragraphs represent objects, actions, and complete tasks in symbolic form.  References to these objects, actions, and tasks can be encoded, stored (memorized), retrieved, and communicated compactly.  Compact storage and retrieval from memory might be advantageous even without communication, and thus apply to less evolved species, i.e., provide survival benefits to precursors to communicating species.  In the paragraphs below we shall discuss brain functions and indications for symbolic encoding schemes.

• We shall use this concept of learning with randomization and selection in explaining the 'learning' of an inner language, and in the 'learning' of an outer language.

In engineering terms we can see this as a method for optimizing by exploring neighbouring (adjacent) traits. By reiterating this method of randomizing about the present location of the trait and finding the 'better' trait variants, we are using local optimization in a search space.

• We shall use this concept to justify why showing the feasibility of a trait or method, and the lack of feasibility of other related traits or methods, increases the probability that evolution will have tried, and may be using, the most successful of the traits and methods that were investigated.
• We shall use this concept in explaining the 'learning' of both inner and outer language.

Most computational (functional simnulation) models of language focus on communication between individuals (what we call outer language). These models investigate language competence and use somewhat later in life, i.e. after the period when the grammar, correct pronounciation, writing and concepts are being learned. There is childhood development work on learning (outer) language, with much of it focused on the correct use of concepts.

We propose a very different approach, based on modelling and social engineering.