Early in the 17th century, Wilhelm von Humboldt observed the discrete nature of what we now call phonemes – the “atoms” of speech, which speakers combine to make words, phrases and sentences. In order to read these utterances, we must be able to hear their phonemic elements in order to recognize their combined forms as verbal. Humans are adept at picking out speech sounds even under very noisy conditions – in other words, isolating sequences of them from the ambient noise, just as a neuron uses a myelin sheath to insulate its signals from the electrochemical storm going on all around – and virtually isolating each phoneme from its neighbors in the sequence, in order to recognize ordered parts of the sonic stream as particular words.
The discreteness and articulation of phonemes is crucial to the functioning of spoken language as a symbol system, and we humans must learn to hear discreteness (as ‘articulated sound’) even when the stream of sound is actually continuous. Recordings of normal spoken language, displayed on an oscilloscope or otherwise ‘objectively’ observed, do not contain gaps of silence between phonemes. This is why ‘motherese,’ the peculiar style of articulation that adults use in speaking to very young children, exaggerates the discreteness of phonemes (see Kuhl et al. in Damasio et al. 2001): in order to acquire language, children have to grasp this discreteness before they can begin to combine the elements of language into words and sentences and thus comprehend (or produce) them.
The discreteness of speech sounds does not contradict the continuity of sound. Likewise there is no contradiction between gradual development and “punctuated equilibrium,” between creation and evolution, or between “sudden” and “gradual” enlightenment.
oh bless the continuous stutter
of the word being made into flesh.— Leonard Cohen, ‘The Window’ (1993, 299)