Phonemes and symbols

The phoneme

We have been studying some of the sounds of English. It is now necessary to consider some fundamental theoretical questions. What do we mean when we use the word "sound"? How do we establish what are the sounds of English, and how do we decide how many there are of them?

When we speak, we produce a continuous stream of sounds. In studying speech we divide this stream into small pieces that we call segments. The word 'man' is pronounced with a first segment m, a second segment a; and a third segment n. It is not always easy to decide on the number of segments. To give a simple example, in the word 'mine' the first segment is m and the last is n, as in the word 'man' discussed above. But should we regard the aI in the middle as one segment or two? We will return to this question.

As well as the question of how we divide speech up into segments, there is the question of how many different sounds (or segment types) there are in English. We introduced the set of vowels found in English. Each of these can be pronounced in many slightly different ways, so that the total range of sounds actually produced by speakers is practically infinite. Yet we feel quite confident in saying that the number of English vowels is not greater than twenty. Why is this? The answer is that if we put one of those twenty in the place of one of the others, we can change the meaning of a word. For example, if we substitute as for e in the word 'bed' we get a different word: 'bad'. But in the case of two slightly different ways of pronouncing what we regard as "the same sound", we usually find that, if we substitute one for the other, a change in the meaning of a word does not result. If we substitute a more open vowel, for example cardinal vowel no. 4 [a] for the Š in the word 'bad', the word is still heard as 'bad'.

The principles involved here may be easier to understand if we look at a similar situation related to the letters of the alphabet that we use in writing English. The letter of the alphabet in writing is a unit which corresponds fairly well to the unit of speech we have been talking about earlier - the segment. In the alphabet we have five letters that are called vowels: 'a', 'e', 'i', 'o', 'u'. If we choose the right context we can show how substituting one letter for another will change meaning. Thus with a letter 'p' before and a letter 't' after the vowel letter, we get the five words spelt 'pat', 'pet', 'pit', 'pot', 'put', each of which has a different meaning. We can do the same with sounds. If we look at the short vowels ɪ, e,  æ, a,  ɒ,  ʊ, for example, we can see how substituting one for another in between the plosives p and t gives us six different words as follows (given in spelling on the left):

'pit'      pɪt

'pet'     pet

'pat'     pæt

'putt'   pΛt

'pot'     pɒt

'put'     pʊt

Let us return to the example of letters of the alphabet. If someone who knew nothing about the alphabet saw these four characters:

'A' 'a' 'ɑ' 'u'

they would not know that to users of the alphabet three of these characters all represent the same letter, while the fourth is a different letter. They would quickly discover, through noticing differences in meaning, that 'u' is a different letter from the first three. What would our illiterate observer discover about these three? They would eventually come to the conclusion about the written characters 'a' and 'a' that the former occurs most often in printed and typed writing while the latter is more common in handwriting, but that if you substitute one for the other it will not cause a difference in meaning. If our observer then examined a lot of typed and printed material they would eventually conclude that a word that began with 'a' when it occurred in the middle of a sentence would begin with 'A', and never with 'a', at the beginning of a sentence. They would also find that names could begin with 'A' but never with 'a'; they would conclude that 'A' and 'a' were different ways of writing the same letter and that a context in which one of them could occur was always a context in which the other could not. As will be explained below, we find similar situations in speech sounds.

If you have not thought about such things before, you may find some difficulty in understanding the ideas that you have just read about. The principal difficulty lies in the fact that what is being talked about in our example of letters is at the same time something abstract (the alphabet, which you cannot see or touch) and something real and concrete (marks on paper). The alphabet is something that its users know; they also know that it has twenty-six letters. But when the alphabet is used to write with, these letters appear on the page in a practically infinite number of different shapes and sizes.

Now we will leave the discussion of letters and the alphabet; these have only been introduced in order to help explain some important general principles. Let us go back to the sounds of speech and see how these principles can be explained. As was said earlier, we can divide speech up into segments, and we can find great variety in the way these segments are made. But just as there is an abstract alphabet as the basis of our writing, so there is an abstract set of units as the basis of our speech. These units are called phonemes, and the complete set of these units is called the phonemic system of the language. The phonemes themselves are abstract, but there are many slightly different ways in which we make the sounds that represent these phonemes, just as there are many ways in which we may make a mark on a piece of paper to represent a particular (abstract) letter of the alphabet.

We find cases where it makes little difference which of two possible ways we choose to pronounce a sound. For example, the b at the beginning of a word such as 'bad' will usually be pronounced with practically no voicing. Sometimes, though, a speaker may produce the b with full voicing, perhaps in speaking very emphatically. If this is done, the sound is still identified as the phoneme b, even though we can hear that it is different in some way. We have in this example two different ways of making b - two different realizations of the phoneme. One can be substituted for the other without changing the meaning.

We also find cases in speech similar to the writing example of capital 'A' and little 'a' (one can only occur where the other cannot). For example, we find that the realization of t in the word 'tea' is aspirated (as are all voiceless plosives when they occur before stressed vowels at the beginning of syllables). In the word 'eat', the realization of t is unaspirated (as are all voiceless plosives when they occur at the end of a syllable and are not followed by a vowel). The aspirated and unaspirated realizations are both recognized as t by English speakers despite their differences. But the aspirated realization will never be found in the place where the unaspirated realization is appropriate, and vice versa. When we find this strict separation of places where particular realizations can occur, we say that the realizations are in complementary distribution. One more technical term needs to be introduced: when we talk about different realizations of phonemes, we sometimes call these realizations allophones. In the last example, we were studying the aspirated and unaspirated allophones of the phoneme t. Usually we do not indicate different allophones when we write symbols to represent sounds.