In the present model we use the term languaging to refer to a general process of representing ideas by sequences of conventionalized signs for purposes of communication. The signs are considered as internal mental programs for the production of patterns of movements of muscle groups (acts) which produce structural information in the environment. The latter we refer to as the environmental, or sometimes simply external, display.
For spoken language, we consider that the signs are internal programs for the complex, coordinated patterns of movements used in speaking (ie., abdomen, chest, and verbal articulators-vocal cords, tongue, lips, etc.). These muscular acts impart characteristic patterns of movements of molecules in the air which may be sensed as structural information in the external display by a listener auder.
Research in speech perception (Liberman, 1970) has suggested that the movements of the articulators, and hence the movements of the molecules in the air, do not faithfully follow the internal articulatory program which the speaker uses to activate the muscle groups involved in speaking. There is a recoding (loss) of information from the formation of internal articulatory programs to the external representation of these programs in speech. That is, the speech signal does not faithfully represent the information in the internal articulatory program (ie., the "plan" for "telling" muscle groups what to do to produce verbal signs). Rather, the speech signal loses information contained in the mental articulatory program and, hence, is said to be an encoded representation of the internal articulatory program which the speaker has formed as a language representation of a conceptualization he wishes to communicate.
Familiar evidence for the lack of agreement between the speech signal and the internal articulatory program is in the lack of ability we have in segmenting the speech stream of a foreigner speaking an unfamiliar language. There is little information in the speech stream to tell us where words begin or end. Thus, such segmenting must be done "in the head" of the language user. Additional, not so familiar, evidence of the code-like nature of speech is the finding that the speech stream does not carry any physical representation of many phonemes which are in fact "heard" by the auder (Liberman, 1970, p.308). For instance, the syllable "da" can be segmented by a speaker/hearer as two phonemes, the /d/ consonant and /a/ vowel. But the physical acoustic stream cannot be so divided. In the physical signal, when the /a/ is removed, what is left is a chirping or gliding sound.