Phonetics and Phonology
Complementary distribution and Free variation
Not all sounds of a language are necessarily distinctive sounds. Compare the English and American pronunciations of "dance". Although there are different sounds in the pair, the meaning does not change. Thus, [a as in barn] and [a as in pat] are not phonemes in this case. We call this phenomenon free variation. The two sounds can be referred to as allophones. These sounds are merely variations in pronunciation of the same phoneme and do not change the meaning of the word. Free variation can be found in various dialects of the same language. In this case, the different pronunciations of words throughout a country do not change the meaning of those words.
Another example of sounds which are not phonemes are those which occur in complementary distribution. This means that where one sound of the pair occurs, the other does not.
Consider the following words with respect to the plosive sounds p,t, and k. Put your hand in front of your mouth and pronounce the words. Do you feel a difference?
pill spill till still kill skill
Did you realize, that there is a burst or puff of air after the /p/ in pill, till, and kill, that is absent in spill, still, and skill? The feature that makes the difference between the plosive sounds in pill, till, kill and spill, still, skill is called aspiration (the period between the release of the closure of a consonant and the start of the vocal cord activity for the vowel that comes after it. This period is usually felt as a puff of air.)
Aspirated and unaspirated allophones are one example of complementary distribution: where the one (e.g. the aspirated p) occurs, the other cannot occur. Aspirated [aspirated p], as you can see in this example, occurs only at the beginning of words. [aspirated p] and [p as in pit] are only allophones of the same phoneme /p/.
The word “complementary” actually refers to the fact that the contexts in which the allophones of a phoneme appear can never be the same and they cover the whole range of possible environments in which the sound can occur (for an analogous situation think of complementary angles in geometry). In other words, in a given context X only a certain allophone will occur, while in another context Y, another allophone is expected to occur and X and Y are the only contexts in which the allophones can occur.
It follows from this that the occurrence of allophones is always predictable since in a certain context we can only expect one and only one realization of the phoneme. In the context of the word pill for example – the voiceless plosive /p/ is followed by a stressed vowel and is in syllable-initial position – we can safely say that the aspirated allophone [ph] will come up. If, on the other hand, p is not syllable-initial and is preceded by s as in spill, we can safely predict that the unaspirated variant of p will occur.
The occurrence of different phonemes is, on the contrary, totally unpredictable since it is the very fundamental characteristic of phonemes that they are contrasted in one and the same context. There is no way in which we can predict therefore that in the context -il we will have pill, nil, chill, fill, gill, Jill, sill, kill, mill, hill, dill or till (the list can continue). Any two words – such as pill and bill, mentioned above, or kill and hill, etc. – that help us discover which sounds have a contrastive value in a given language are said to form, just to remind you, a minimal pair.
The following criteria must be met by the two words in order that they form a minimal pair: they should have the same number of sounds, and these sounds should be identical, with the only exception of the contrasting sound that should be distributed in the same context in both words; the words must also have different meanings.