Not only does language have both hit sounds as part of its repertoire, but, like nature, it treats the unreleased “t” sound and the released “t” sound as the same phoneme. This is remarkable, because they are temporal opposites: one is like a little explosion, the other like a little antiexplosion. One can imagine, as a thought experiment, that people could have ended up with a language that treats these two distinct “t” sounds as two distinct phonemes, rather than two instances of a single one. In light of the auditory structure of nature, however, it is not at all mysterious: any given hit can have two very different sounds, and language carves at nature’s joints.
In light of the two sounds hits make, there is a simple kind of sound we can make, but that language never includes as a phoneme: “beep,” like an electronic beep or like Road Runner. A beep consists of a sudden start of a tone, and then a sudden stop. Beeps might, at first glance, seem to be a candidate for a fundamental constituent of communicating by sound: what could be simpler, or more “raw,” than a beep? However, although our first intuitions tell us that beeps are simple, in physics they are not. In the real world of physical events among objects, beeps can only happen when there is a hit (the abrupt start to the beep), a ring that follows (the beep’s tone), and a second hit, this one a dampening one (the abrupt beep ending). A “simple” beep can’t happen in everyday physics unless three simple constituent events occur. And we find that in languages as welclass="underline" there are no beeplike phonemes. To make a beep sound in language requires one to first say a plosive of the released kind, then a (nonwiggly) sonorant, and finally an unreleased plosive . . . just like when we say the word “beep.”
Hesitant Hits
Bouncing a basketball could hardly be a simpler event. A bounce is just a hit, followed by a ring. And as we discussed earlier, the sound is a sudden explosion of many frequencies at the initiation of the hit, followed by a more tonal sound with a timbre due to the periodic vibrations of the basketball and floor. Although hits seem simple, they become complicated when viewed in super slow motion. After the ball first touches the ground, the ball begins to compress, a bit like a spring. After compression, the ball then decompresses as it rises on its upward bounce. Although these ball compressions and decompressions are typically very fast, they are not instantaneous: the physical changes that occur during a hit occur over an extended period of time, albeit short. What happens during this short period of time depends on the nature of the objects involved.
One of the most important acoustical observations about collisions is that ringing doesn’t tend to occur until the collision is entirely finished. There are several reasons why this is so. First, the ground rings less during the collision because even though the ground has already been struck, the ball’s contact with the ground dampens the ground’s vibrations. Similarly, the ground’s contact with the ball dampens the ball’s vibrations. Second, during the ball’s compression, its shape is continually varying, and so any vibrations it is undergoing are changing in their timbre and pitch very quickly, far more quickly than the ring-wiggles we discussed earlier. In fact, the vibration changes occur at a time scale so short that any rings that do occur during the collision will not sound like rings at all. Third, during the period of the collision when the ball is not yet at maximum compression, the ball is continually hitting new parts of the ground. This is because, as the ball compresses, the ball’s footprint on the ground keeps enlarging, which means that new parts of the ball continually come into contact with the ground. In fact, even if the surface area of contact never enlarges, the mass in parts of the ball continues to descend during the ball’s compression, providing further impetus upon the surface area of contact. Because the compression period is filled with many little hits, any ringing occurring during compression will have a tendency to be drowned out by the little hits.
For several converging reasons, then, the ringing that occurs after a hit doesn’t tend to begin until the compressions and decompressions are over. For the basketball, the ringing occurs most vigorously when the ball rebounds back into the air. There is a simple lesson from these super slow motion observations: there is often a gap between the time of the start of a collision and the start of the ringing.
What determines the length of these hit-to-ring gaps? When your basketball is blown up fully, and the ground is firm, then the time duration of the contact with the floor is very short, and so the gap between the start of the hit and the ring is very small. However, when the ball is fairly flat—low in pressure—it spends more time interacting with the ground. Bouncing a ball on soft dirt would also lead to more ground time (see Figure 6). Figure 7a shows the sound waveform signal of a book falling onto a crumpled piece of paper—producing similar acoustics (in the relevant respects here) to those of a ball dropped on soft dirt—and one can see a hit-to-ring gap that is larger than that for the same book falling directly onto the table. For the flat basketball, then, the gap between hit and ring is larger than that for the properly blown-up ball.
Figure 6. (a) A rigid hit (i.e., involving rigid objects) rebounds—and rings—with little delay after the initial collision. (b) A nonrigid hit takes some time before rebounding and ringing. These physical distinctions are similar to the voiced and unvoiced plosives.
The key difference between the high-pressure ball and the flat ball—and the difference between the book falling on a solid desk versus crumpled paper—is that the former is more rigid than the latter. The more rigid the objects in a collision, the shorter the compression period, and the shorter the gap between the initial hit and the ring. The high-pressure ball is not only more rigid than the flat ball, but also more elastic. More elastic objects regain their original shape and kinetic energy after decompression, lose less energy to heat during compression, and tend to have shorter gaps. Also, if an object breaks, cracks, or fractures as it hits—a kind of nonrigidity and inelasticity—the gap is longer.
Therefore, although some hits ring with effectively no delay, other kinds of hits take their time before ringing. Hits can be hesitant, and the delay between hit and ring is highly informative because it tells us about the rigidities of the objects involved. Our auditory systems understand this information very welclass="underline" they have been designed by evolution to possess mechanisms for sensing this gap and thus for perceiving the rigidity of the objects involved in events.
Because our auditory systems are evolutionarily primed to notice these hit-to-ring delays, we expect that languages should have come to harness this capability, so that plosives may be distinguished on the basis of such hit-to-ring delays. That is, we would expect that plosive phonemes will have as part of their identity a characteristic gap between the initial explosive sound and the subsequent sonorant. Language does, indeed, pay homage to the hit-ring gaps in nature, in the form of voiced and unvoiced plosives. Voiced plosives are like “b,” “g,” and “d,” and in these cases the sonorant sound following them occurs with negligible delay (Figure 7b, left). They even sound bouncy—“boing,” “bob,” and “bounce”—like a properly inflated basketball. Unvoiced plosives are like “p,” “k,” and “t,” and in these cases there is a significant delay after the plosive and before the sonorant sound begins, a delay called the voice onset time (Figure 7b, right). (Try saying “pa,” and listen for when your voice kicks in.) In English we have short voice onset times and long ones, corresponding to voiced and unvoiced plosives, respectively. Some other languages have plosives with voice onset times in between those found in English.