 In the following, I will apply the main principles of morphological analysis to an example from present-day English. I will confine the analysis to British English and receive pronunciation as the standard accent. Furthermore, I will show that even though we are familiar with the present-day English orthography, that the basis for a morphological analysis must be phonology. The examples I've chosen concern the formation of number, that is singular versus plural. And even though present-day English uses non-concatenative operations in this morphological process, such as woman-women or mouse-mice, I will confine myself to the concatenative operation, that is plural formation by means of affixation. Here are my examples presented orthographically. Dog-dogs, cat-cats, horse-horses, box-boxes, echo-echos, thief-thieves, ox-oxen. Now on the basis of an orthographical representation, we would have to postulate the following plural. Well, let me say endings. We have something like a simple S. We have something like ES. Or here. Well, and we have something like EN, if we compare the plural with the singular. So using orthography, it seems as if there are different conditions under which you have to add a simple S versus ES, and then maybe as a special case, EN. That this is not so, and that there is some different condition at work here, can only be shown if you provide a morphological analysis based on phonology. Now it becomes clear that dog and dogs and echo-echos, for example, well, that they do not have different endings, but that in both cases a voiced alveolar fricative is added to the singular form in order to generate the plural. And also horses, where you would add a simple S in orthography, and boxes, where you would have to add ES in orthography, have the same affix phonologically. Let's perform the next step and isolate the affixes. Well, in this case, we have suffixes only. Now we have three variants. We have the variant where a voiced alveolar fricative is added, where a voiceless alveolar fricative is added, and where a more complex suffix is added. The EN, well, that may be a special case. So what sort of alternation have we got? We have six stems that remain unchanged, where the singular stem form is identical with the plural stem form. And then we have one stem where the stem changes. In the singular, we have a voiceless final fricative in thief, and in the plural, the final element is a voiced labudental fricative. Well, and then we have four affixes, where one, the one in ox, oxen can be excluded, or it is relatively rare, because it only attaches to some selected stems. I've listed some of them here. Ox, oxen, child, children, with a stem change, and an additional alveolar approximate r, and brother, brethren, a very specific word. And that's it. Now two of these involve a stem change, so this is a very special case. So let's concentrate on the situation with the voiced alveolar fricative z, the s and the is. So let's look at their distribution. Well, the voiced alveolar fricative occurs after all those stems that are voiced before the word boundaries. Dog, echo, echo, the final vowel is voiced, of course, and thief. The voiceless alveolar fricative attaches to all those stem forms where the final element is voiceless, as in cat. Now this would also apply to horse or box because the final element is voiceless. However, here we have a special condition, and that condition is ranked highest. So we have to check that first. If any of these phonemes stands before the stem boundary, then we have to add is. Well, with this in mind, we can now apply a principle that has long been known. This principle was first introduced by Charles Hockett in the mid 1940s and is referred to as normalization or preliminary normalization. So what is it? Well, the idea is that in the case of morphological alternation, we first of all define a general morph that involves a set of alternatives as a default. Well, of course, in our case, this must be the voiced variant, the simple voiced variant, because the majority of the environments where this one occurs are nouns that end on a voiced phoneme, and there are more voiced phonemes in English than voices one, consider the fact that 50% or even more of the consonants are voiced, nasals are always voiced, and vowels are of course voiced, so voiced consonants occur in words like echo or dog. So the default is clearly the voiced alveolar fricative. Now with the default, we can now connect a set of alternation rules, where the default has to be changed. So rule 1 to rule n, and perhaps then we define a context where no change occurs. Well, these alternation rules can be subdivided as follows. First of all, we have phonological alternation, and we've already said that the default changes into is, if any of these phonemes stands before the word boundary, and it changes into the voiceless alveolar fricative, if any other voiceless element constitutes the last segment before the morph boundary, in words such as cat-cats. Then we have what morphologists call lexical conditioning. Well, in principle, this means that the information about the choice of a plural is stored in our lexicon, that is, we have to remember it. So we simply have to remember that the plural form of ox is oxen, of child is children, of brother, special type of brother is brethren, and so on. Well, and finally we have a special condition, which we call morphological alternation or morphological conditioning, where the morphological category determines the choice of alomorph. So here we have to postulate a special case for the category plural, where the stem changes, well, let's mark it here, from voiceless ones as in thief, life, and cloth into voiced final stem elements, thief, life, cloth. Now, in cloth clothes there is also a vowel change involved. Well, and this has to be contrasted with the environment of the genitive singular, where no stem change occurs, so the genitive is still phonologically conditioned. It is thieves, lives, and cloths. Note that in this case, in the case of morphological conditioning, we also have to remember the various words, because it is on the one hand thief thieves, but on the other it is chief and chiefs, and it is cloth clothes, but moth moths. So in other words, the morphological condition must be combined with a lexical condition. So let's now summarize. Morphological analysis must be based on a phonemic representation. In our case, echoes and dogs would have been different in orthographical terms, but they are identical in terms of a phonemic analysis. Using the phonemic analysis, we first have to isolate the morph, so that's always something we have to do first. And then we set up a normalized form in the case of variation. In our case, a normalized affix. Well, and if there are alternation rules, well, we have to define them very precisely on the basis of the conditions under which they operate.