However, the problem in viewing the possessor as the specifier of the DP is that this would predict that possessors can appear in front of a determiner, when in actual fact possessors and determiners seem to be in complementary distribution:
|c||*Johnís the book|
If X-bar theory is correct however, this observation cannot be taken to show what it seems to: i.e. that the specifier and the determiner sit in the same structural position. Words and phrases cannot be in complementary distribution as they cannot appear in the same positions in a phrase. Some other element must appear in complementary distribution with the determiner in (39c), not the possessor. But what?
To answer this question we first have to consider another property of the possessor. Pre-nominally, possessors are marked by the element Ďísí. What is this morpheme? Some have suggested that it is the marker of genitive Case born by the possessor. However, if this is a Case marker, it is a very strange one for at least two reasons. First it is a Case marker in a language which does not usually mark Case on its nominal elements. English only normally marks Case on its pronouns and noun forms are typically invariant no matter what Case position they occupy. Yet if we take Ďísí to be a marker of genitive Case we have to assume that nominals are marked for this Case. The other strange thing about this morpheme seen as a marker of Case is that it does not behave anything like a Case morpheme in any other language. Note that it does not attach itself to the noun, but to the last element in the whole DP:
|b||that manís book|
|c||the man that I told you aboutís book|
|d||the man that you metís book|
This behaviour is consistent with the claim that this morpheme does not attach itself to any word, but to the whole phrase. Note this is not the way that Case morphemes behave in other languages. In Hungarian, for example, the accusative case morpheme Ėt is attached to the noun inside the DP, not to the last element in the DP:
|(41)||a||egy kťpet Marirůl|
|b||*egy kťp Marirůlt|
|(42)||a||the manís going|
|b||the man that I told you aboutís going|
|c||the man that you metís going|
|(43)||a||the maníll do it|
|b||the man that I told you aboutíll do it|
|c||the man that you metíll do it|
The contracted auxiliary attaches itself to the end of the subject in much the same way that the possessive morpheme attaches itself to the end of the possessor. The difference is, however, that with the contracted auxiliary there is an uncontracted form:
|(44)||a||the man is going|
|b||the man that I told you about is going|
|c||the man that you met is going|
|(45)||a||the man will do it|
|b||the man that I told you about will do it|
|c||the man that you met will do it|
Presumably, what happens when the auxiliary verb contracts is that it undergoes some process which attaches it to the subject. Very likely this is not a syntactic movement, but a phonological process which takes place after the structure has been constructed.
Evidence in favour of this comes from the comparison of auxiliary contraction and negative contraction, which does involve a syntactic movement. When the negative element not contracts, it sticks itself onto the auxiliary verb in front of it:
|(46)||a||I will not talk|
|b||I wo-nít talk|
If the auxiliary then moves, the contracted negative is taken along. Thus when the auxiliary inverts with the subject in certain questions, the negative also inverts and cannot be left stranded behind the subject:
|(48)||a||D-structure:||Theodore thinks [who Ėíll win]|
Indeed, auxiliaries can contract onto a subject that has moved to the position in front of the auxiliary, suggesting that this contraction takes place after movement:
|(49)||a||D-structure:||[DP e] will seem [this man to disappear]|
The point is, however, that there is a process which takes an independent word and sticks it to the phrase immediately in front of it. If this is what is going on with the possessive construction in English, then the Ďísí morpheme must originate as an independent word which sits in a position immediately following the possessor. Given that the word position immediately following the possessor is the determiner, we conclude that the morpheme Ďísí must be a determiner. Unlike auxiliaries, this determiner has no uncontracted form and so we never see it occupying the D position. We thus have the following analysis:
If this analysis can be maintained, we now know why possessors and determiners are in complementary distribution; in fact they are not in complementary distribution but the possessive determiner Ďísí is in complementary distribution with other determiners, as we would expect.
There is one drawback to this analysis however, which concerns pronoun possessors:
Presumably as these pronouns have the same function as possessors, they sit in the same position: specifier of the DP:
Note that these pronouns have a special genitive form, which demonstrates that this position is one to which genitive Case is assigned. Thus, even if the Ďísí morpheme is not the marker of genitive Case, DPs which sit in the specifier of a DP have this Case. The problem is that with pronoun possessors the possessive determiner (the Ďísí morpheme) does not appear. This is rather puzzling, especially if this morpheme is nothing to do with genitive Case.
There are a number of possible solutions we might suppose. One is to assume that the possessive determiner is present with pronoun possessors, but remains unpronounced. This is supported by the fact that, as with all other possessors, no other determiner can appear with a pronoun possessor:
|(53)||a||*my the house|
|b||*her a travel permit|
If nothing is in the determiner position, these observations would be hard to account for. We then have to assume that for some reason when the possessor is a pronoun, the possessive determiner is unpronounced and when it is a non-pronoun it gets pronounced as Ďísí. As far as I know there is no explanation as to why this should be and so it remains as a descriptive statement at present.
An alternative would be to claim that the possessive determiner is always unpronounced and hence that the Ďísí morpheme is not a realisation of this determiner at all. Instead it is a marker of possession, which pronouns do not need as they have a genitive form to demonstrate their status as possessors. The problem with this is that it is tantamount to claiming that the Ďísí morpheme really is a Case morpheme after all, despite it not behaving like one.
A third possibility would be to claim that the reason why pronoun possessor are in complementary distribution with all other determiners, including the possessive determiner, is because they are determiners sitting in the head position. From this perspective, the structure of the DP with a possessive pronoun would be:
This solves all the previous distribution problems, but places the pronominal possessor in a different structural position to all other possessors, which makes their similar interpretations difficult to account for. Moreover, (54) is not likely to be the correct analysis for semantic reasons. The reference of the possessor is obviously different to the reference of the whole DP: the pronoun refers to a person (i.e. you) whereas the DP refers to a mistake. But if the pronoun is the head of the DP, how could it have a reference that differed from the DP? It seems that there is no perfect solution to these problems from our present understanding of the internal organisation of the DP and we will therefore have to wait for further developments to make progress in this matter.