Tone terracing


Tone terracing is a type of phonetic downdrift, where the high or mid tones, but not the low tone, shift downward in pitch after certain other tones. The result is that a tone may be realized at a certain pitch over a short stretch of speech shifts downward, then continues at its new level and then shifts downward again until the end of the prosodic contour is reached, at which point the pitch resets. A graph of the change in pitch over time of a particular tone resembles a terrace.
Since the pitch of the low tone remains more-or-less constant at the lower end of the speaker's vocal range, and the other tones shift downward, the difference in their pitches narrows, eventually obscuring the tones altogether. Pitch reset is then required if the tone system is to continue to function.
Tone terracing is particularly common in the languages of West Africa, where typically only the low tone causes downstep. However, a somewhat more intricate system is found in the Twi language of Ghana. Twi has three phonemic tones: high, mid, and low. A word, and therefore a prosodic chunk of speech, may only start with a high tone or a low tone on its first syllable. As in many languages, a low tone starts out and remains at the bottom of the speaker's range. After a low tone, a subsequent high tone is downstepped.
However, a phonetic downstep occurs between any two adjacent mid tones as well. In fact, a high tone is defined as any tone that is at the same pitch as a preceding high or mid tone; a mid tone will always be lower in pitch than a preceding high or mid tone. The result is that every instance of a mid or low tone shifts the upper end of the pitch range downward, until all pitches are reset at the end of the prosodic melody.
Table 1. The phonetic terracing effect in Twi of a series of mid tones.
Table 2. The phonetic terracing effect in Twi of an alternating series of high and mid tones.
Table 3. The phonetic terracing effect in Twi of an alternating series of high and low tones.
From tables 2 and 3, it can be imagined that the tone sequences high-low-high and high-mid-high may be difficult for a non-native speaker to distinguish.