Publications of Dr. Martin Rothenberg:
The Breath-Stream Dynamics of Simple-Released-Plosive Production

5.  Air Flow Resistance at the Articulatory Constriction

A basic problem in the study of breath-stream dynamics is to relate the variations of articulatory resistance (Ra) to the underlying physical and physiological parameters, according to the various possible places of articulation and their modifications. For each place of articulation and modification, the eventual goal would be to describe the different phonetically significant closing and opening gestures. Given the additional pertinent physiological variables of an individual production, for example, those that affect the pressure which tends to burst the articulatory closure open, it would be desirable to relate each phonetically different articulatory gesture to the resulting variation of articulatory resistance. However, we have considered it beyond the scope of this work to discuss these complex and as yet poorly defined problems. In this present chapter we will primarily consider the air flow resistance as an independently varying parameter and briefly discuss direct techniques for its measurement. The restrictions imposed by the dynamic properties of the articulatory mechanism on the duration of the closure, though of obvious importance in plosive production, are considered beyond the scope of this present study. Thus, closure duration will not be considered at all in this chapter, though it will be mentioned in Chapter 8.

Unfortunately, there appears to have been no previous study in which direct measurements were made of the variation in air flow resistance at the articulators during the opening or closing phase of a plosive consonant. Such an experiment would entail the simultaneous measurement of the pressure variation across the constriction at the place of articulation and of the volume velocity through the constriction. [See the Measurement of Air Flow in Speech (Journal of Speech and Hearing Research, Vol. 20, March 1977, pp155-176) for later measurements.]

One source of information, however, is the study of air flow by ISSHIKI and RINGEL (1964) in which a number of simultaneous recordings of supraglottal pressure and air flow were made. While these investigators do not report any attempt to determine the variation in articulatory resistance from their recordings, they show both air flow and pressure records for a sequence of two CVC syllables, one of which, [paph], is reproduced above as Figure 3.4.2. Though the scales in the record are in some respects inadequate for accurate measurements of the transients during the articulatory release, a number of semi-quantitative observations can be made which are indicative of the type of data desired in a study of the variation of articulatory resistance in plosive production :

(1) In the initial plosive in Figure 3.4.2, the variation in articulatory resistance Ra during the closing movement can be computed (assuming negligible nasal air flow) by dividing the supraglottal pressure by the flow rate. Such a computation, starting from the point in time at which the supraglottal pressure begins to rise appreciably, reveals that the articulatory resistance rises approximately exponentially to about 70W in a period of about 120 msec. During this interval, the resistance doubles approximately every 20 msec. Shortly after the flow resistance reaches 70W, the air flow drops to zero rather sharply, creating a transient in both the air flow and pressure traces. The articulatory resistance appears to rise from 70W to ¥ in less than 10 or 15 msec.

The above observations have the following implications in the study of breath-stream dynamics: (a) If the total flow resistance in the glottal-supraglottal air stream, excepting that at the place of articulation, is low compared to 70W, as with a preceding fricative or sibilant, the time constant of the increase in Ra can be considered to be about 40 msec (the time required for a four-fold increase in Ra). (b) If the total additional flow resistance in the air stream is of the same magnitude or higher than 70W, as with a preceding voiced sound, the time constant of the increase in Ra must be considered to be less than 15 msec (less than the time required for Ra to increase from 70W to ¥). This faster time constant is illustrated in the closing gesture for the final plosive, during which the volume velocity appears to decay to zero within one or two glottal cycles.

(2) In the final plosive shown in Figure 3.4.2, most of the variation of articulatory resistance takes place in the first 40 or 50 msec after the release. After that period, the volume velocity is approximately proportional to the pressure across the articulators. (The supraglottal pressure shown is roughly equal to the transarticulator pressure.) The factor of proportionality is the articulatory resistance, and approaches approximately 2W. This value indicates a not-too-open final lip position, as could be expected in a terminal release.

(3) At approximately 10 to 15 msec after the release for the final plosive, the articulatory resistance stops decreasing, increases slightly, then continues its descent. This oscillation occurs at a mean articulatory resistance of approximately 8.9 cm H2O / 0.64 liter/sec or 14W

The oscillation of the articulatory orifice in a bilabial plosive, though not clearly shown in Figure 3.4.2 due to the time scale, has been noted in the measurements of orifice area made by FUJIMURA from photographs (1961). A rough sketch of what the articulatory resistance function of the final plosive of Figure 3.4.2 might look like has been presented in Figure 2.5 above.

A careful study of the pressure and volume velocity records of the initial plosive in Figure 3.4.2 has indicated to the writer that there is here also an oscillation of articulatory resistance which occurs about 10 msec after the release and has a mean resistance of about 2.8 cm H2O / 0.65 liter/sec or 4.3W

Primarily due to the interference of the time lines, this conclusion is far from certain. However, the figure of 4.3W is consistent with the assumption that the articulators are moving to a position for [a] which is more open than the position which they assume following the release of the final plosive.

(4) As would be expected, both the post-release volume velocity and the pressure at the articulators are of different orders of magnitude for aspirated and unaspirated plosives. In particular, the peak volume velocity in an aspirated plosive apparently can be at least as high as 2 liter/sec. (Though the peak of the volume velocity curve was not shown in the original reference, a rough extrapolation can be made.)

(5) If the rapid changes of pressure and volume velocity immediately after the time of release of a plosive are to be recorded with any degree of confidence, the pressure and volume velocity recording systems should have a response time of less than about 2 msec, at least in the case of a bilabial plosive. It is likely that the same requirements are posed for measurements of plosive articulations made with the equally mobile tip of the tongue.

Little of a quantitative nature is known concerning the variation of articulatory airflow resistance for plosives in which the closure is made with surfaces of the tongue other than the tip. It seems to this author that a closure made with the blade of the tongue would not tend to be exploded open in the same way that a typical bilabial closure might be, and that there would not be an oscillatory resistance function as noted for the bilabial case. However, there appears to be no experimental evidence at present to corroborate this hypothesis.

Of possible interest in this regard is the observation that palatal and velar released plosives often do not have a single clearly defined transient marking the release in the acoustic signal, but instead have a short sequence of rather irregularly spaced transients. An obvious explanation is that the tongue remains in close proximity. to the palate or velum for some considerable period of time, and that the release is composed of a sequence of bubbles or ripples travelling through the moist interspace as the surfaces separate.

It is not the task of this chapter to evaluate such hypotheses, but merely to indicate that there is a need for more data in this area, and that good measurements of the simultaneous variation of airflow and trans-articulator pressure might be very helpful and not discouragingly difficult to instrument.

Chapter 4 Passive Variations of Subglottal Pressure in Plosive Production
Last Chapter
  Chapter 6 Change of Glottal Adjustment in Plosive Production
Next Chapter


Table of Contents

E-mail Dr. Rothenberg