Consonants

Stops

Stops are characterized on a spectrographic display by silence (stop gap) or obvious signal weakening. This is the acoustic interval corresponding to the articulatory occlusion, and it varies in duration depending on the prosodic condition. Because the vocal tract is obstructed, little or no energy is produced. This is followed by a burst of energy, as the closure is released.

There are two dimensions that need to be carefully analyzed in distinguishing stop sounds. One of them is the separation of fortis from lenis (or, as more commonly known, voiceless from voiced). Several indicators can help us to identify the stops with reference to this dimension. These are:

(a) Duration of the stop gap, i.e. the silent period during the closure phase. Specifically, /p, t, k/ show longer closure duration than /b, d, g/.

(b) Presence of a voice bar, i.e. a dark bar found at low frequencies (generally below 250 Hz) in a spectrogram. Except in intervocalic position, this is one of the least reliable identifying characteristics of the stop sounds. As we saw earlier, voicing is frequently absent in English /b, d, g/ in initial and final positions. Thus, this feature can reliably be used to separate /p, t, k/ from /b, d, g/ only in intervocalic positions; /b, d, g/ will show the voice bar that is indicative of voicing, whereas /p, t, k/ will not.

(c) Release burst indicated by a strong vertical spike. In general, we observe a stronger spike for /p, t, k/ than for /b, d, g/. One qualification is in order here for the final position: as discussed earlier, final stops in English are normally unreleased. In such cases, we will not observe any release burst. Velar stops are said to be more prone to non-release than bilabials or alveolars.

(d) Duration of the previous vowel. This refers to positions other than the initial position, where there is no vowel before the stop. In other positions, vowels and/or sonorant consonants are longer before /b, d, g/ than before /p, t, k/.

(e) Aspiration. A VOT of more than approximately 30 ms shows up as short frication noise (scattered marks after the release) before vowel formants begin, in initial /p, t, k/ of a stressed syllable.

The other dimension in the identification of stops is the place of articulation. While the above criteria are useful in separating voiced (lenis) from voiceless (fortis), we do not gather much information with respect to the place of articulation. In other words, what should we look for to distinguish a /p/ or a /b/ from a /t/ or a /d/, or from a /k/ or a /g/? The following are helpful in this respect:

(a) Formant transitions. Formant transitions (shifts) in CV sequences reflect changes in vocal tract shaping during stop-to-vowel transition. Vowel to-stop (VC) transition is the opposite of the former. Frequency of the first formant of the vowel increases when stops are at the beginning of a syllable (CV), and falls when they are at the end (VC); this is the case for all stops. While no information for the place of articulation can be gathered from F₁ transitions, F₂ and F₃ transitions can tell us a great deal in this respect. In CV situations, bilabials show upward movement of both F₂ and F₃ (more reliable before front vowels than before back vowels); the mirror image situation is revealed in a VC sequence. For a velar stop, F₂ and F₃ are close together just after the stop is formed (more reliable before back vowels; before front vowels, F₃ and F₄ have common origin). There will be a downward transition to a vowel with a high F₂ . In cases of a VC, there is a narrowing (coming together) of the second and the third formants. The case of alveolar stops is probably the least straightforward; while we observe a flat transition to a vowel with mid or high F₂ , there is a downward transition to a vowel with low F₂ in CV cases. In VC cases, there is very little movement; small downward movement of the second formant and small upward movement of the third formant are expected.

(b) Locus (starting frequency). This is the point of origin to which the transition of the second formant appears to be pointing. F₂ and F₃ loci are described as an inherent property of the place of articulation for obstruents and nasals, although formant loci are not necessarily visible in the spectrogram. For a /b/, this is around 600–800 Hz; for a /d/, it is 1,800 Hz. To identify a /g/, two loci, at 3,000 Hz and 1,300 Hz, are generally noted.

(c) Release burst. Bilabials are identified by bursts with a center frequency lower than the F₂ of the vowel (below 2,000 Hz). They show a pattern that is diffuse and weak. Alveolar bursts generally have a center frequency that is higher than the F₂ of the vowel (above 2,000 Hz). The pattern is diffuse and strong. As for velars, they display a compact and strong pattern, and the bursts have a center frequency approximating to the F₂ of the vowel.

(d) VOT (/p, t, k/ only). The length of time from the release of a stop until voicing begins for the following segment presents consistent differences depending on the place of articulation of the fortis stop. The time interval includes the spike (the sound that is produced by the separation of the articulators), a short frication noise after the spike, and the aspiration following it. As we move the place of articulation from the front to the back of the mouth, stops tend to have greater VOT. Although this is a rather effective acoustic property for classifying the place of articulation, it should be remembered that it is just a tendency, and that it may be possible to find an alveolar voiceless stop with a longer VOT than a velar voiceless stop. The VOT for a syllable-initial voiceless stop varies somewhat systematically depending on the place of articulation and other segmental context.

If we examine the spectrograms of the words speak, peak, keen, and cake (figures 2 and 3), we can see these points clearly. In speak, the /p/ is not at the beginning of a stressed syllable, since it is preceded by a /s/; thus we do not have aspiration in the voiceless bilabial stop. The words peak, keen, and cake all have voiceless stops at the beginning of a stressed syllable, and thus are aspirated. However, as one can easily see, the degree of the voice lag (aspiration) is different in each case; shorter in peak, longer in cake, and the longest in keen. The reasons for these differences are as follows. Among the three, peak has a bilabial stop whose expected lag is lower than that of the other two words that start with a velar stop. The difference between the two words that start with velar stops (cake and keen) is due to the type of vowel that follows the voiceless stop. The aspiration in keen is longer than in cake, because in the former /k/ is followed by a high vowel (a vowel with lower sonority), while in the latter the stop is followed by a lower vowel (a vowel with higher sonority).

Summarizing all the above, the examples confirm the statements made earlier: that voiceless stops are not aspirated if they are not at the beginning of a stressed syllable; when they are aspirated, the degree of aspiration is greater as we move the place of articulation from front to back (i.e. bilabial, to alveolar, and to velar). When a stop with a given place of articulation is followed by vowels of different height (different degrees of sonority), the amount of lag varies; a stop before a high vowel has greater aspiration than when it is before a low vowel. Flaps are different from their alveolar stop counterparts /t/ and /d/. In the production of stops, the vocal tract is blocked for about 50 ms, but in the flap, which is produced by a rapid throw of the tongue against the alveolar ridge, the duration is very short, about 20–30 ms; as such it is the fastest consonant.

مواضيع ذات صلة

Fricatives

Diphthongs

Vowels

Acoustics of Vowels and Consonants

English Vowels SUMMARY

Full (Strong) Forms versus Reduced (Weak) Forms of Function Words

Full Vowels–Reduced Vowels

Non-US Varieties

Diphthongs

High back vowels

Mid back vowels

Back Vowels Low back