PhD summary

Summary of Messum (2007), 'The role of imitation in learning to pronounce'

In my UCL PhD thesis (downloadable from the Articles page), I addressed fundamental issues in two areas of phonetics.

In Part 1, I examined stress-accent and three so-called ‘timing’ phenomena in English and other West Germanic languages. I argued that our current understanding of pre-fortis clipping, ‘rhythm’ and the lengths of tense and lax vowels as genuinely time-based cannot be correct. The reasons for this include developmental data that contradicts this notion and the lack of a plausible learning mechanism by which children would acquire these timing patterns at the very young ages that some of them begin to appear.

Instead, I developed an alternative account of the underlying nature of these phenomena and how they develop. As part of this, I conceived a new role for speech breathing in speech development, and showed how phoneticians have been in serious error in neglecting this aspect of speech production. It has not been appreciated that speech breathing in children occurs against a completely different mechanical backdrop from speech breathing in adults. (The tissue of the chest wall, in particular, is much more compliant in children.) In addition, the aerodynamic factors of speech do not scale consistently between adult and child. The models of speech breathing in each population are thus significantly different.

Working from these fundamental principles, it seems that sentence stress in children learning West Germanic languages is implemented by respiratory system activity, and that this determines the timing of segments in certain contexts, the timing of larger prosodic groupings and other aspects of the final output. Natural extensions of my account explain a number of other longstanding problems in phonetics, including voice onset time variability, ‘syllable cut’ phonology, and P-centres (perceptual-centres).

In Part 2, I looked at how children learn to pronounce speech sounds. I argued that phonetics has again been in serious error, this time in basing its theoretical superstructure on the unexamined assumption that children learn speech sounds by imitation. There is, in fact, no evidence for this, and what evidence there is points against the idea. I developed a hypothesis based on a proposal originally made by Caleb Gattegno (1963), that children learn speech sounds through a mirroring interaction with their caregivers; mirroring being a well-attested developmental mechanism for other aspects of psychological development and one for which there is empirical evidence of the necessary interactions between mothers and their children.

This account of speech development is superior to the conventional one in many ways, and its implications for phonetics would be far reaching. It would resolve basic contradictions between phonetic theory and data from neuroscience, explain hitherto anomalous data on speech shadowing and allow us to develop an integrated picture of the roles of production and perception in speech development.

Its significance goes even further than this: it would resolve the most fundamental, long-standing issue in speech, that of the underlying nature of the phenomena. Under this proposal, neither Stetson (speech as gestures made audible) nor Bell and most current theorists (speech as primarily an acoustic phenomenon) are correct: instead, speech is a genuinely perceptuo-motor phenomenon, with a correspondingly integrated underlying representation in the brain. Our present assumption of a ‘simple’ learning mechanism for speech sounds (imitation) has lead to a very confused picture; a slightly more complex social learning mechanism (mirroring) makes the data coherent.

I should add that although my ideas are not proven, they are entirely consistent with the experimental data in speech and speech development. I have taken the time to visit and present to many of the leading researchers in the fields concerned. In no case have these experts raised any objections to the theoretical side of my proposals or pointed out data that would tell against them.