What is the underlying form of the English plural marker? The plural 's' has three variants: the /s/ in 'cats', the /z/ in 'dogs' and 'pies' and the /Iz/ in 'boxes'. These variants occur in specific environments. The /s/ occurs after voiceless consonants, the /z/ after voiced consonants and vowels, and the /Iz/ variant after consonants and affricates which have greater acoustic noise and turbulence than other consonants. If we take the case of the words 'judges', 'matches', 'wishes' and 'races', we have examples of affricates in the first two words and strident fricatives in the last two.
Many textbooks of linguistics argue that the underlying form of the plural marker in English is /z/. They make this claim on the basis of wider distribution. The voiced alveolar fricative is present not only in /z/ but also in /Iz/. However, is this really correct? First of all, it is debatable as to whether or not we should even claim that there is an underlying form. Do we want to say that at one time in the historical development of English there was only one form and from that form other forms developed? A counterclaim is that each form is contextually dependent and as a result the notion of wider distribution is invalid.
However, if we must choose an underlying form, we can also present many arguments in favour of the allomorph /s/. The first is that in forms such as 'snows' and 'plays', it is true that either a voiceless or voiced alveolar fricative can occur. We have words such as place' and 'race' to prove this. Nevertheless, because the word-final vowel is voiced, we can argue that a voiced fricative could be the result of assimilation. Perhaps the underlying form was once voiceless and later became voiced as the result of an assimilation process.
Other arguments in favour of /s/ as underlying are orthography. If the plural marker had originally been pronounced as a voiced fricative, would this not be reflected in the orthography? Markedness theory also favours /s/. It is well-known that voiceless fricatives are less marked than marked ones and that in word-final position many languages only have voiceless fricatives such as German and Dutch. English, though, is a Germanic language with many voiced fricatives in word-final position as in 'knives', 'boys' and 'dogs'.
Other languages such as Spanish and German have a voiceless fricative in the plural: i.e., Spanish has 'gatos' (cats) and German has Autos (cars). This also provides evidence to suggest that the underlying plura marker could be /s/.
A problem for the claim that the underlying marker is /s/, though, is forms such as 'knives' and 'lives'. The Swedish, Norwegian and Danish words 'kniv' and 'liv' indicate that 'knive' and 'live' may be underlying forms in which the word-final consonant devoiced. If this is the case, /z/ appears to be underlying. But it is also possible that 'lives' and 'knives' are the result of a progressive voicing assimilation, (live + s > live + z; knife + s > knife + z), and the devoicing that resulted in 'life' and 'knife' occurred at a different stage in the development of the English language. This goes against the claim of the Neogrammarians that sound change is always regular, but unfortunately, despite the wishes of many linguists, it does not appear that language is always so simple.
Many linguists claim that the underlying plural allomorph in English is /z/. However, it is also possible to claim that all the allomorphs are contextually dependent and thus occurred simultaneously in their own specific environments. If an underlying form must be chosen, it is also possible to claim that this underlying allomorph should be /s/ and not /z/.