 In our e-lecture, the phoneme 1, we define the phoneme physically as an abstract head term of a family whose members, the alophones, satisfied two central criteria. They must be phonetically similar and they occur in complementary distribution. Functionally, the phoneme can be defined as the smallest linguistic unit capable of distinguishing meaning. Thus, there must be some sort of discovery procedure for phonemes. So, we will first look at this discovery procedure and we'll then use two examples to extend our physical view by the phenomenon of free variation. Now, let's look at the discovery procedures first. The most well-known method of phoneme discovery is the method of setting up minimal pairs. That is, words whose sound structure is identical except that they differ in one speech sound. Normally, minimal pairs are thus very short words. For example, the two words over here represented by these phonemes ban versus pan. Let us illustrate this discovery procedure, the minimal pair procedure using the word Paul as an example. As you can see, Paul has been described narrowly here with the diacritic superscript H to signal aspiration. And the final alveolar lateral which is velarized. Now, let's look at some possibilities. Well, the left column over here wants us to define minimal pairs by replacing the initial element. So, let's take the following words. Paul may be tall, mall. So, you see the use of different phones in this initial position immediately changes the meaning of the words. Paul is something different, tall and mall. But what about this one here, where we say Paul. So, I'm just dropping the aspiration component, Paul versus Paul. Well, you would say this sounds a little bit strange but you can understand me. So, we have all the blue variants as distinctive phonemes whereas the green one doesn't change the meaning. So, the aspirated bilabial plosive and the non-aspirated ones must be members of the same phoneme. Let's play the same game with the internal vowel. So, here are some variants. For example, peel would be one or a diphthong here, pale. Maybe if you wish in British English, pearl. Well, and then, well, let's use Paul again but this time we raise the vowel slightly and instead of saying Paul we would say Paul. Paul versus Paul. Paul, Paul, not pale but pole. So, a slight raising of the vowel. Well, this wouldn't change the meaning really. We would sound a little bit awkward and we would run into difficulties with the contrastive word pole instead of hall. So, we have to be careful but nevertheless I would say the raising of the or to or would not change the meaning. And finally, let's take the last segment. So, here we would have something like pause for example or in British English without a post-vocalic R we would get something like pork. We could get something like port again without a post-vocalic R. And what about this one? If we take the palatalized variant and instead of saying Paul we're saying Paul. Well, the meaning wouldn't change but we would sound awkward. So, in other words, the green ones are variants of one phoneme whereas here we have different phonemes. So, this is how the procedure of setting up minimal pairs works. It is a suitable method of phoneme discovery. Once we have found the phonemes within a language we can define the actual variants that is their alophones. Let's take another example. Here it is. It looks a little bit strange but what I want to do is I want to focus on the bilabel plosives in this example. Well, let's first of all look at the actual realizations. The bilabel plosive in Paul is aspirated. Paul. The bilabel plosive in stop is incomplete. It involves a sort of absence of closure. So, we do not say stop but stop. And this little diacritic here is responsible or denotes the lack of aspiration, the incomplete plosion. Well, and the internal one in spoke is neither incomplete nor aspirated. Well, if you don't believe me that this is an unaspirated per, listen to the following example which I took from the glossary of the virtual linguistics campus. Now here we have several words with a very similar phonemic structure. They start with an alveolar, voiceless alveolar fricative sir and then a plosive follows. Now by gradually cutting off the initial fricative portion in these words, in a word like spin for example, you would see that the second phoneme loses its identity. What is normally regarded as a voiceless aspirated plosive consonant seems to become unaspirated. If not voiced, the more portions of the S are cut off. Well, let's listen. Spin, spin, spin, spin, spin, spin. Let's stop here and now close your eyes and listen. Spin, spin. Well, you see aspiration is gone. So it's no longer a true voiceless plosive. Okay, let's return to our example. So we have three possible variants of a bilabial plosive. Let's move them down a little bit. They are phonetically similar. They're all bilabial plosives. Well, and they are distributed complementarily. So the first one occurs before vowels as in Paul. The second one occurs after this phoneme after an alveolar fricative, a voiceless alveolar fricative. And the final one occurs before a morph boundary. Well, this is a typical example of a phoneme where we have three alophones. Well, and the symbol chosen for this phoneme is of course the standard Roman character. Let us now look at another example where we focus on the realization of orthographic R. That R is a phoneme in present day English can quickly be shown. In our sentence, Ron carried a present. You can contrast Ron with Don or with gone and immediately you get different meanings. So the R must be a phoneme in its own right or take present in which case you could contrast it with a pleasant. Okay, so the R must be a phoneme. Now let's look at possible variants. Well, in the first case, we have an alveolar approximate in Ron. We also have an alveolar approximate in carried and we have a slightly devoiced alveolar approximate in present. Well, the contexts can be quickly defined. The devoiced variant occurs after p as in present, t as in trick or k as in crack. Whereas these two can be found in any other position. Well, that's not the end of the story because especially in British English there is a variant. In Ron, we would always say Ron, we would use the alveolar approximate but between vowels. Here in this context, we could optionally use the alveolar flap as in carry. So we could say Ron carried a present. This phenomenon is referred to as free variation. Free variation is normally bound to particular contexts. But sometimes free variation may not just be a contextual variant, a distributional variant, but it's also a matter of social or dialectal variation. For example, the use of R and interestingly it seems to be the R that is especially affected by this phenomenon of free variation across languages. Here are some examples. Well, this is the cognate comparison component in the language index of the virtual linguistics campus. Now, you can select a number of words which we have recorded. Let's take the word for three because that contains an R in many languages. And let's now first of all move down to German. So here is an example from Franconian. Drei, clearly an alveolar trill. Let's now compare that with a speaker from Berlin. Drei, where the speaker clearly uses the uveolar trill. And just a few kilometers further down. Drei. Another speaker uses the voiced velafricative. And if you go down south and select a speaker from southern Germany, you get the alveolar flap. Let's now play the same game with Italian. In Italian, you would normally have the alveolar trill as a standard variant. However, there are speakers who simply don't use it. Let's compare two speakers from Brescia. Drei. Clearly an alveolar trill. And from the same town. Drei. A speaker with a variant, the velafricative, the voiced velafricative. Our final example comes from English. Now, my first speaker is someone from West Virginia. Drei. Who clearly uses the standard variant of American English spoken in the United States. The retroflex approximate three. Now, if we compare this speaker with a speaker from Canada from Ontario, then we get something like this. Drei. Clearly, not a retroflex approximate anymore, but an alveolar approximate. Now, the situation is different if we go to India and listen to a speaker from Thane Marashta. Drei. Clearly, apart from the vocalic differences, an alveolar flap. So, these variants then show that free variation may be an effect of different dialects. And also, it may be sometimes an aspect of social class membership where you try to use a particular variant of the realization of this R. As you can find out in social linguistic studies, especially the ones carried out by William Lobov in the 1970s. Well, let us summarize. The phoneme is the smallest linguistic unit that can distinguish the meaning of words or morphs. The phoneme can be defined as a family of speech sounds with an abstract head term. So, here would be the abstract head term. And the symbol used for the abstract head term is normally a Roman character. If we run out of Roman characters, then we take other characters taken from other alphabetic writing systems. The family members, the alophones, should be phonetically similar. They should occur in complementary distribution. And sometimes, and this was illustrated in this e-lecture, they have variants that stand in free variation. These variants may be contextually defined, but they can also be dialectal or social variants.