The linear utterance is the final product of successive cipherings :

  1. The phonological ciphering :
The phonemes  with which the speaker builds the signifiers ("signifiants") belong to a closed system of oppositive units. In Saussure's signs, signified and signifier are closely linked. Apart from the phonemic system stricto sensu there are rules of prosody which  apply to the linear string (rhythm and intonation). Phonology seems to me the best introduction to  linguistics:  the systematic patterning of the phonemes as combinations of minimal distinctive features  is a good stepping-stone to the systematicity of grammar.
  1. The lexical ciphering :
The utterer has inherited a lexical stock which is by no means a series of labels attributed to the objects of the extralinguistic world . Nothing will be said here about  the possibility of a "mentalese" underlying the structuring of utterances ( see Steven Pinker 's point of view in his "Language Instinct").
  1. The syntactic ciphering :
The processing of the utterance consists in a series of deep abstract operations. On the grammatical level, the utterer does not operate with ready-made chunks but plays with the potentialities of his grammar. In doing so  he takes into account the situation of enunciation, the verbal context  and the nature of the intended message.
The morphology of the lexical elements ( grammatical number of nouns, agreement, verbal endings etc.) and the order of their appearance in the final linear chain, as well as the tracers of the structuring operations (free or bound morphemes such as
a/the, this/that, to/-ing, do, un/le, ŕ/de, bien, etc.)  are the surface  tokens of the ciphering operations executed by the utterer-architect. The result of this complex processing (Humboldt's ergon) represents an algebra which mirrors in a revealing way the encoding of the utterance (and not the extralinguistic world , hence the impossibility of a direct assignment of the meaning ). One can understand easily  the totally artefactual character of an analysis which tries to assign meaning to the successive elements of the linear string. However this is exactly what has always been done and what is still currently being done in  grammar. This atomistic approach  has produced  catastrophic results  both in the conception of grammar-books and in … the brains of the learners ! How can one  account for the use of a BE+ING construction by focussing on the verb ALONE ? What image can one give of the deep function of the French Imparfait  if one does not see what sort of predication is being signalled by the -AIT metaoperator?
Languages offer varying degrees of
readability ("lisibilité). For example in English DO reveals a major metalinguistic operation of predication ; in German, the order of elements in subordinate clauses conveys precious information  on the organization of the sentence as a whole; in French the ŕ/de microsystem  can almost be deciphered at surface level. What must be kept in mind is the fact that a particular grammatical point can be a cryptotype in one language and  an almost transparent marker in another. This explains why it is impossible to describe a language in isolation . Contrastivity is  thus an absolute necessity in linguistic analysis.

The linear utterance is not the message

Biographical Notes     Linguistic Research: Books and Papers     
The Linguistic Credo of Metaoperational Theory     
The Genesis of Metaoperational Grammar
The Double Keyboard of Grammars     
The 10 Components of the Metaoperational Model