Piers Messum - Research website
How do young children learn to pronounce their mother tongue (L1)?
Not (at the start) by imitation ...
These are my child speech research pages. For pronunciation teaching, visit Pronunciation Science Ltd.
I am interested in how young children learn the pronunciation of L1 (their first language), and in particular:
the qualities of its speech sounds
its timing patterns
My accounts of the two learning processes resolve some fundamental problems in speech development and phonetics. They point the way to more effective pronunciation teaching for older learners.
Almost everyone believes that children learn speech sounds and speech timing phenomena by imitation. Not by imitation in the sense of simple mimicry, of course, but by young children identifying and then copying the important characteristics of what they hear.
This belief is widespread for several reasons, including these:
that it appears to be ‘common sense’. The information is there in the speech signal and replication occurs, so why not by copying?
that children certainly do learn to pronounce words by imitation, in one sense of the word 'imitation'.
But learning the sequence of speech sounds that make up an individual word - which is what an experienced speaker does in order to learn the pronunciation of a new word - is not the same as learning speech sound pronunciations. These are the actions needed to reproduce speech sounds themselves, which are then available for use in the production of words. Speech sounds are part of the system of a language. Individual words make use of that system.
Solving the 'correspondence problem' for speech sounds:
Possibility #1, acoustic matching
The conventional assumption: the child does all the work.
(Yellow arrows indicate the activities undertaken by each participant.)
In the picture above, the child is a social isolate, struggling on his own to make sense of the speech sound 'data' presented to him, and to match it in some way (despite having a vocal tract that cannot produce sounds with all the same properties as adults). In the picture below, his caregivers are involved with him in games of vocal play, as attested in child development studies everywhere, and (without knowing that this is what they are doing) they help to launch him into the pronunciation of L1 speech.
Possibility #2, 'mirrored' equivalence
Reformulation by caregivers: the perceptual judgement is made by the expert, adult speaker; the equivalence between his actions and the sounds she 'mirrors' back to him is deduced by the infant.
The mechanism is well attested in practice. It shares the work, and solves some longstanding problems in speech.
Turning to the temporal organisation of speech, the fact that we can measure a reliable timing effect (say, a relative shortening of one vowel compared to another one in a particular consonantal context) does not mean that this was the speaker's intention. William James called the confusing of observational experience from the actual experience of the subject the "psychologist's fallacy." Timing effects can be epiphenomenal, a by-product of some other mechanism.
Phoneticians and others do not generally consider the possibility that there might be alternatives to simple copying to explain how children end up pronouncing words like the people around them. But the proposition that all the basic elements of pronunciation are learnt by imitation is no more than a belief. There is no evidence for ‘imitative’ accounts of either the learning of speech sounds or the learning of the temporal phonetic phenomena that characterise particular languages.
In my thesis I laid out the problems with the conventional view, including the 'anomalies' which are currently disregarded. I then described alternative mechanisms, showing:
how the reflection (reformulation) of a child’s utterances by his mother can lead to the emergence of speech sounds (as considered in the pictures above), and
how the aerodynamics and respiratory physiology of speech in a child-size body can lead to behaviour that we have wrongly construed to be timing-based. (I.e. changes in timing occur, but not intentionally.)
In neither case is the child's production modelled on what he hears.
In comparison to what is currently believed, my proposals are more plausible and more coherent.