8. Summary and Discussion
The gross time and space variations of air pressure and volume velocity within the vocal mechanism during speech have been referred to as the breath-stream dynamics of speech. The present study is of the breath-stream dynamics of the class of speech sounds referred to here as the simple-released-plosives. These speech sounds are defined as those released stops with an egressive, pulmonic pressure and a single articulatory closure, in addition to a possible velopharyngeal closure. The simple-released-plosives present a level of intermediate complexity in the study of the breath-stream dynamics of speech.
An attempt is made in this study to develop and show the relations between a minimally redundant set of physiological parameters which are complete enough to describe most of the phonetically significant variations in the breath-stream dynamics of the simple-released-plosives. The parameters are presented within the framework of a linear, lumped-parameter model, using electrical circuit terminology and symbols.
Though some new data are presented, the major effort is toward interpreting the findings of the many studies in physiology and physiological phonetics of recent years which are relevant to the breath-stream dynamics of plosive consonant production. Thus, the present study is primarily an attempt to make explicit a set of assumptions that are consistent with the present state of knowledge of the speech mechanism and the phonetic nature of speech. It is hoped that such a formulation will help focus attention on unsolved problems and suggest directions for future research.
The production of a simple-released-plosive may require rapid coordinated movements of the oral, articulatory, laryngeal, pharyngeal, velar, and respiratory structures. In this study the movements of these structures have been represented (in circuit terminology) by three resistance elements (Ra, Rg and Rn), a current source (Ie), and a voltage source (Em; or in a more detailed model Em-ab, Em-th and Em-ab,th). A model was developed for the dynamic interrelations between these 'control' parameters in the production of simple-released-plosives. The model was designed to represent the average (non-acoustic) variations of pressure and air flow (the breath-stream dynamics). The model relates the control parameters to certain other linear, lumped-parameter elements representing the less controllable parameters of the vocal mechanism.
These other, 'non-control' parameters involved in the model are the compliance of the air in the supraglottal cavity (Co), the properties of the walls of the supraglottal cavity (Rw, Lw and Cw), the compliance of the air in the subglottal airways (Csg), the subglottal air flow resistance (Rfl), and the subglottal (respiratory) tissue parameters (Rt, Lt and Ct; or in a more detailed model Rab, Rth, Lab, Lth, Cab, Cth and Cab, th).
The above set of control parameters is in two respects unnecessarily redundant for the description of the phonetically distinct variations in the breath-stream dynamics of simple-released-plosives:
(1) There are a number of conditions under which the actual movement of interest is partially determined by the action of the other control parameters. In such an event the independent (and therefore less redundant) parameter may be thought of as the underlying adjustment of the structure of interest. For example, the respiratory movements (as the movements of the thorax) in the production of a plosive may be highly dependent on the variation of glottal-supraglottal air flow resistance. Therefore in this work the control of the respiratory movements (and of subglottal air pressure) is represented by a muscle activation parameter (Em-t, or Em-ab and Em-th) that is directly related to the changes in innervation of the muscles of respiration.
The effect on subglottal pressure of changes in glottal-supraglottal resistance is derived by simulation. A comparison with the few data in the literature indicates that the order-of-magnitude numbers presented in this study for Rfl and/or Rt may be too large by as much as a factor of two. These resistances and many of the other subglottal parameters are not known with any precision for the conditions involved in normal speech.
Also treated in some detail in this study is the relationship of the 'movements or actions' of the glottis in plosive production to the underlying 'adjustments' of the glottis. It is an obvious consequence of the generally accepted aerodynamic theory of voicing that a vibratory action of the true vocal folds (voicing) requires not only a suitable adjustment of the glottis but also a sufficient transglottal pressure and air flow. In plosive production these latter (breath-stream) parameters are determined to a large extent by the other, non-laryngeal control parameters (Ra, Rn, Ie and Em). It is shown in this study that any appreciable voicing during the period of articulatory closure requires either an incompletevelopharyngeal closure or a passive or active expansion of the supraglottal cavity. It is likely that each of these three mechanisms is significant in some language.
However, the notion that the action of the glottis is not always an independent parameter in speech does not appear to be reflected in the current phonetic terminology for plosives. It is strongly suggested by the results of this study that a simple-released-plosive should be classified phonetically as 'voiced', not on the presence or absence of voicing during the period of articulatory closure, but on the presence or absence of a voiced glottal adjustment during this period (or under some conditions a breathy-voiced or tightly-voiced adjustment). When there is a purposive mechanism present, one effect of which is to substantially increase or decrease the duration or strength of voicing, it is suggested that the term 'voiced' be modified. For example, if there is a purposive expansion of the pharynx, the term 'voiced-pharyngealized' could be usedl. If the voicing is supported by an incomplete velopharyngeal closure, the term 'voiced-nasalized' might be appropriate.
(2) Any specification of an adjustment or movement by a time-varying parameter is redundant unless the specification includes all the dynamic constraints inherent in the adjustment or movement. In this study the dynamic constraints on a control parameter in plosive production are described in terms of the time constant for the variation of the parameter that results from a 'ballistic' movement that is either 'unidirectional' (a step change between two quasi-steady-states) or 'cyclic' (returning to the same or a similar quasi-steady-state). It is suggested that a 'ballistic' movement can be defined as a maximally fast movement that is still adequately controllable for the particular conditions of speech.
From order-of-magnitude values for the subglottal parameters, it is shown that the inertive constraints due to the masses of the respiratory structures are relatively small compared to the physi-ological constraints in the (active) change of respiratory muscle tension. It appears that the time constant for a unidirectional ballistic increase of subglottal pressure during the period of closure of a plosive at a moderate rate of speech is about 150 msec. The time constant for the corresponding increase in the control parameter Em is probably slightly less, or about 130 msec.
The results of this study indicate that an important movement of the glottis in plosive production is a ballistic cyclic opening movement, voiced-open-voiced (or voiced-whispered-voiced). This author considers it likely that most, if not all spoken natural languages use a gesture of this nature for some intervocalic simple-released-plosive, with the timing of the glottal movement having a relatively fixed relation to the opening and closing movements at the articulators. A shift in the timing between the articulatory movements and the cyclic glottal opening movement may be used in the production of a phonemic contrast or, apparently more frequently, in the production of the difference between a single intervocalic plosive phoneme and a geminated pair. The time required for a ballistic cyclic opening movement of the glottis at a moderate rate of speech appears to be about 120 to 150 msec.
The dynamic constraints in the closing and opening movements of the articulators were not considered quantitatively. Though some data are available in the literature concerning articulatory movements, it was decided not to consider this aspect of plosive production in detail because of the great variety in the types of movements possible for the various places of articulation and their modifications.
Also important in plosive production in some languages are the dynamic constraints on movements of the velum. Though some observations are presented in this study concerning the functioning of the velum in plosive production, no quantitative data were obtainable with the methods employed.
A major goal in this study was to delineate some of the unsolved problems in the study of the breath-stream-dynamics of speech. It was found that more information is desirable in the following general areas:
(1) the mechanism of muscle contraction;
(2) the patterns of motoneuron activity in rapid, coordinated, learned movements;
(3) the contraction properties of the various muscles involved in speech production;
(4) the subglottal tissue and air flow parameters involved in speech production;
(5) the air flow resistance of the glottis in states other than normal voicing;
(6) the action of the glottis and the air flow resistance during a voiced closure, when the transglottal air pressure varies greatly during each glottal cycle;
(7) the possible role of a tightly-voiced glottal adjustment in plosive production;
(8) the action of the true vocal folds immediately after the plosive release for various glottal adjustments, with and without prerelease voicing;
(9) the patterns of articulatory movement for the various places of articulation and their modifications; the resulting variations of air flow resistance; the conditions under which the supraglottal pressure becomes a significant factor in producing the articulatory release;
(10) the ability of the speech mechanism to synchronize the articulatory and glottal movements; the variety of contrasting timing patterns to be found in natural languages;
(11) the passive properties of the walls of the supraglottal cavity for the various plosive articulations;
(12) the extent to which active supraglottal cavity expansion is used to sustain voicing in phone types normally considered plosives (not implosives);
(13) the movements of the velum in nasalized plosives; the relations, if any, between movements of the velum and changes of glottal adjustment in phonoaspirated plosives;
(14) the possible effect of an opening or closing movement of the glottis on the rate of oscillation of the true vocal folds;
(15) the relative invariance of the glottal adjustment (as compared with the invariance of the glottal action) in voiced plosives;
(16) a measure of phonetic distance along each of the relatively independent control parameters of simple-released-plosive production.