Voice onset time

As stated earlier, a stop articulation consists of a closure formed by the two articulators followed by an abrupt release of this closure. We will look at the production of stop sounds and the timing of vocal cord vibration, which is relevant for voiced, voiceless, aspirated, and unaspirated

distinctions. The differences for these various kinds of stops can be explained by the time difference between the release of the stop closure and the beginning of vocal cord vibration. This timing relationship is known as the ‘voice onset time’ (hereafter VOT). Figure 1 represents the different stop productions in the VOT continuum.

If the voicing starts before the release (i.e. during the closure period), as in the case of lines (a) and (b), then the situation is described as having ‘voice lead’ and given a negative VOT value (in milliseconds; ms). Line (a) represents a fully voiced stop; we have vocal cord vibration throughout the closure, which continues after the release. The /b, d, g/ sounds of Romance languages are said to be typical examples of fully voiced stops.

Not all voiced stops are produced in this fashion. In some languages, English and other Germanic languages included, /b, d, g/ are subject to a certain amount of loss of voicing (‘partially devoiced’) during their production. Line (b) in figure 1 represents this configuration; the voicing starts some time into the closure stage and continues into the following vowel (the mirror image of this is seen in final voiced stops).

If, on the other hand, the voicing starts after the release of the stop closure, then it is said to have a ‘voice lag’ and is described with a positive VOT value (in milliseconds; ms). Cross-linguistically, the amount of lag may be significant; while a lag greater than 30 ms results in stops that are called ‘aspirated’ (or ‘long lag’), a shorter voice lag or voicing simultaneous with release results in stops that are known as ‘unaspirated’. Lines (c) and (d) show these two possibilities. In neither case do we have vocal cord vibration during the stop closure (thus ‘voiceless’). The difference between the two cases lies in the point at which the voicing starts with respect to the moment of release. In line (c), the vocal cord vibration is simultaneous with the stop release; the VOT is zero and we have a ‘voiceless unaspirated stop’. The voiceless stops of Romance languages are given as examples for this.

In line (d) the lag is longer than the 30 ms threshold, and the resulting sound is a ‘voiceless aspirated stop’. The diacritic used for aspiration is a small raised [^h] to the top right of the stop (e.g. [p^h]). English initial [p^h, t^h, k^h] sounds are produced in this way and we hear the resulting short burst before the buzz of voicing in the vowel. The degree of aspiration may be different in different languages. For example, while English voiceless stops are slightly aspirated, their counterparts in languages such as Mandarin, Thai, and Scots Gaelic are strongly aspirated.

In some languages (e.g. Hindi of India, Sindhi of Pakistan and India), the possibilities go beyond the three types of stops (voiced, voiceless unaspirated, voiceless aspirated) we have discussed, with the addition of the so-called ‘voiced aspirated stops’. These stops have, after the release of the stop closure, a period of breathy voice (murmur) before the regular voicing starts for the following segment. Thus we get the following four-way voicing distinction in Hindi:

[tal] “beat”; [t^hal] “plate”; [dal] “lentil”; [d^hal] “knife”