Coarticulation : A universal phonetic phenomenon with roots in deep time

Coarticulation is a universal feature of spoken languages. Many decades of experimental phonetic research have produced a large literature on the topic. However, despite many important contributions, we still lack an answer to the perhaps most fundamental question about coarticulation: Where does it come from? In keeping with the Frame/Content theory of speech evolution (MacNeilage 2008) our analysis subsumes speech movements in the class of continuous cyclic motor behaviors (swimming, walking, breathing and chewing) and views them as sharing the discrete positional control seen across species in ‘precision walking’ (Grillner 2006) and reaching (Georgopoulos & Grillner 1989). In speech the well-known conservatism of evolution is evident both in its syllabic organization which is built on the oscillatory motion of the mandible and its use of discrete spatial targets –in the production of phonemes. Once this organizational framework is assumed, a natural explanation of coarticulatory overlap presents itself. The overlap arises from the fact that the responses of articulatory structures to discrete segmental goals are slower than the rate at which the open and close states of the syllabic jaw cycle occur. Explaining coarticulatory overlap When we speak, our motor system coordinates a large number of neuro-muscular components. The movement between two consecutive phonemes is rarely a single one-parameter trajectory. The kinematics of a CV syllable is better pictured as a time chart specifying a long list of actions to be performed by articulatory, phonatory and respiratory structures. Even the simplest utterance is a multichannel event. An example makes that evident. Consider the syllable [ku] spoken in isolation. There is no lip activity specified for the stop. Nevertheless, we find that the lip rounding for [u] is in progress during the closure, the articulatory movements for [k] overlapping or being coarticulated with those for [u]. The elementary fact highlighted here is that coarticulation is manifested in a temporal overlap between any two channels recruited by different phonemes. The often cited representation of this fact is the schematic used by Joos in his classic monograph on acoustic phonetics (1948). It shows the beginning of the phrase Wo ist ein Hotel? represented as a series of ‘innervation waves’ overlapping in time. Figure 1. Coarticulatory overlap illustrated by Joos’s classic “Overlapping innervation waves” for the German phrase ‘Wo ist ein Hotel?’ Why do the movements for the adjacent phonemes have to overlap? Could things be otherwise? For one thing, reaching articulatory goals for a phonetic segment takes time. For all articulators to be in position for [u] shortly after the release of [k], the rounding and other movements must be initiated well in advance. As suggested in Joos’s diagram every segment gives rise a wax-and-wane pattern, a period of anticipation followed by a de-activation phase. Second, different channels show different temporal properties. Along a continuum from TMH QPSR Vol. 51