Two types of obstruent

Obstruents (plosives, affricates, and fricatives) come in pairs in English. These pairs are listed in the following charts.

Traditionally the consonants in group 1 are referred to as voiceless, those in group 2 as voiced. This is certainly what the symbols used to transcribe these consonants suggest. However the difference between these pairs is not always that the first one is voiceless, while the second one is voiced. For this reason we will use the label fortis for the obstruents in group 1 and lenis for the obstruents in group 2. This section gives a somewhat simplified overview of how the fortis–lenis contrast is manifested in various positions within the word. Interestingly, the difference between a fortis and a lenis consonant is not always the same.

voicingIf a sound is voiced, the vocal folds are vibrating. You can watch this in slow motion in the following video.

This vibration is easy to maintain when the air is emerging relatively freely from the lungs. In obstruents this is not so. Plosives and affricates are different from fricatives in that the former contain an occlusion, a phase during their articulation when the flow of air emerging from the lungs is totally blocked. In fricatives the air is continuously escaping because there is a narrow gap between the articulators. But even in fricatives the airflow is significantly slowed down because of the narrowness of the gap. Obstruents with an occlusion are called plosives, those without are fricatives.

VOTVOT abbreviates voice onset time, ie the time vocal fold vibration begins relative to the explosion in a plosive (or affricate). This explosion is called the release of the plosive. If voicing begins about when the explosion takes place, VOT is zero. If it begins earlier, we have negative VOT, if later, VOT is positive. (More on this in Wikipedia.) This makes it possible to distinguish three types of plosive: voiced with negative VOT (vocal folds are already vibrating before the release of the plosive), voiceless unaspirated with zero VOT (vocal folds begin to vibrate at the moment of the release), and aspirated with positive VOT (vocal folds begin to vibrate significantly later than the release).

voiceless sonorantsSonorant consonants (nasals — m n ŋ — and approximants — l r j w) differ from obstruents in that the air coming from the lungs can flow relatively freely in them. As a result, the vocal folds are spontaneously vibrating in sonorants, ie sonorants are usually voiced. However, vocal fold vibration may be deliberately inhibited even during the pronunciation of sonorants by pulling the vocal folds apart. This is called aspiration. If VOT is positive in a plosive, the first part of the following sound, a sonorant, is voiceless, since vocal fold vibration does not begin at the release of the plosive. A voiceless vowel, for example, is perceived simply as h. Therefore aspiration is often marked as a h following the plosive: pin phɪn, print phrɪnt, sometimes raised to show that it is part of the preceding plosive, not an independent segment: p^hɪn, p^hrɪnt. Alternatively, we could also write the voiceless version of the following vowel or of the consonant to indicate aspiration: pɪ̥ɪn, pr̥ɪnt.^»A little circle shows that the sound it accompanies is voiceless. Thus aspiration means that the voicelessness of a plosive does not end at the release of that plosive, but extends over the following sonorant segment as well.

aspirated plosivesAt the beginning of a word and at the beginning of a stressed syllable plosives in English are distinguished not by voicing (ie negative vs zero VOT), but by aspiration (ie zero vs positive VOT). Thus the words pin pɪn and bin bɪn are different, because the p is aspirated and the b is not aspirated. The difference is not that the p is voiceless but the b is voiced, as the symbols used to transcribe these words suggest. That is, the pronunciation of pin is phɪn, that of bin is pɪn. (Recall, the h in pin marks the voicelessness of the first part of the vowel ɪ, VOT is positive, voicing begins after the release of the p. In bin VOT is zero, voicing begins at the release of the plosive, therefore the following vowel is voiced right from its beginning.)

Plosives occurring between two sonorants (ie vowels or sonorant consonants) are either fully voiced: eg rubber, amber, rebel, abridge; or voiceless: eg rapper, ample, repel, appraise. At the beginning of a stressed syllable, a voiceless plosive is aspirated: repel rəphɛ́l, appraise əphrɛ́jz. So the b of rebel, abridge and the p of repel, appraise are not only distinguished by the first two being voiced (negative VOT) and the second two voiceless, but also by the lenis plosives being unaspirated and the fortis ones aspirated (positive VOT).

s+plosivePlosives that follow a fortis fricative — within a word this is usually s, and sometimes f, but not other fricatives — are voiceless, but not aspirated. So the plosives of the following words are not aspirated: special spɛ́ʃəl, explain əksplɛ́jn, caftan kaftán. Now we have seen that a plosive before a stressed vowel which is unaspirated is not fortis, but lenis. So we are forced to say that these words contain lenis plosives after the fricatives, ie they are mistranscribed by dictionaries: special sbɛ́ʃəl, explain əksblɛ́jn, caftan kafdán are perhaps more appropriate representations of these words. In any case, fortis and lenis obstruents do not contrast in this position, their contrast is neutralized after fricatives.

word finallyWord final lenis obstruents lose much of their voicing. Word final fortis obstruents are not aspirated: if aspiration is the voicelessness of the following segment, a plosive not followed by anything can hardly be aspirated. What distinguishes a word-final fortis and lenis obstruent is the length of the preceding vowel. If we compare bad and bat, we see that the vowel of bad is significantly longer than that of bat. The same is true in the case of fricatives: the most prominent difference between phase fɛ́jz and face fɛ́js or the verb use jʉwz and the noun use jʉws is in the length of their vowels. This is also the case if a sonorant consonant precedes the obstruent: tend vs tent, tens vs tense, killed vs kilt, cold vs colt.

The following chart summarizes the obstruent allophones in English. Only the labial pairs are given for plosives (p b) and fricatives (f v), the other pairs behave in the same way. (V́ means a stressed vowel, V̊ means an unstressed vowel, sonorant consonants may occur between the vowel and the obstruent.)