One of the most important points to take into account if we want to pronounce vowels correctly is their length. We know that there are long vowels, which in phonetics are marked by a colon after the corresponding symbol (/ɑː/, /ɜː/, /ɔː/, etc.), and short vowels (/ɪ/, /ʊ/, /ə/, etc.). And we should also remember that the phoneme /æ/ has a sort of special status because, even though it’s included in the short vowel group, it’s the longest of the short ones. However, as is often the case in English, that’s not all there is to know. We have to learn about a phenomenon called pre-fortis clipping which changes these notions quite a lot.
The pre-fortis clipping process arises from the fact that the length of a vowel is strongly determined by the voicing of the consonant that comes after it (or by the absence of any consonant, if this is the case). The term fortis is equivalent to voiceless and clipping stands for shortening. So, what this convoluted expression means is that when a stressed vowel is followed by a voiceless consonant within the same syllable, the length of that vowel is considerably reduced. This is especially noticeable in the case of long vowels, which are shortened up to half their length.
Listen to some examples and note how clearly this difference can be seen:
Long vowels
Long vowel followed by voiced consonant or pause Long vowel followed by voiceless consonant
As you can see, vowels followed by a voiced consonant or a pause keep their normal length, whereas those coming before a voiceless consonant become much shorter.
Diphthongs
This process applies to diphthongs as well:
Diphthong followed by voiced consonant or pause Diphthong followed by voicelss consonant
employ /emˈplɔɪ/
exploit /eksˈplɔɪt/
Short vowels
Now, let’s see how pre-fortis clipping affects short vowels too, although the difference is not so obvious:
Short vowel followed by voiced consonant or pause Short vowel followed by voicelss consonant
Consonant clusters
Another interesting fact to remember is that, whenever more than one consonant follow the vowel, the clipping also occurs provided that there is a voiceless one within the same syllable .
Vowel followed by more than one consonant Vowel followed by more than one consonant
Nouns and verbs
There are also pairs of verbs and nouns which are actually the same word, the only difference being that the verb ends in a voiced consonant while the noun ends in its voiceless counterpart. This may be reflected in the spelling (devise-device, believe-belief) or not (abuse, house). And here too, in the case of voiceless final consonants (nouns), the preceding vowel is shortened by the pre-fortis clipping phenomenon.
Verbs Nouns
abuse /əˈbjuːz/
abuse /əˈbjuːs/
devise /dɪˈvaɪz/
device /dɪˈvaɪs/
excuse /ɪkˈskjuːz/
excuse /ɪkˈskjuːs/
believe /bɪˈliːv/
belief /bɪˈliːf/
So, how shall we distinguish hard from heart if both d and t sound more or less the same, you may well ask? The answer is in the length of the vowel, because although voiced consonants are devoiced the length of the preceding vowel remains intact. So, hard and heart are not distinguished by the final d or t, as they sound virtually the same, but by the length of the vowel /ɑː/, which is longer in hard because the /d/, even if devoiced here, is in origin a voiced consonant. This is exactly what native speakers do.