6.1 The structure of IP
Inflectional elements are word level categories such as will, can, may, must, etc. as well as to, ed (and its numerous irregular manifestations) and s. In chapter 1 we argued that these all belong to one category, ‘Inflection’ (I) because of their complementary distribution:
|(1)||a||*Mike might will see the doctor|
|b||*Bill will to go to work|
|c||*Cathy can watches TV.|
In all of the sentences in (1) there are two inflectional elements and each time this produces an ungrammaticality. Therefore not only can we conclude that each of these elements belongs to the same category, but that there is only one position for this category in each clause.
We also suggested in chapter 1 that inflectional elements take verbal complements on the observation that they are always followed by a VP (or perhaps a vP, depending on the properties of the verb). From an X-bar point of view, this suggests that inflections are to be treated as heads as only a head takes a complement. If this is right, then we predict that there will be a phrase that the inflection heads; an IP:
Note that this is a fundamentally different type of structure than the VP that we investigated in the previous chapter. In the VP, apart from the main verb itself, light verbs, both Θ-assigning and aspectual types, select verbal complements and project a verbal phrase. Hence one light verb can take a complement headed by another and a complex VP can be built. Inflections, however, take verbal complements but project an IP. This accounts straightforwardly for why there can only be one inflection per clause as there can only be one IP per clause.
What is the nature of the IP and what else does it contain? Just as inflections are always followed by the VP, they are also typically preceded by the subject in its surface position, though as we pointed out the subject originates inside the VP (vP) at D-structure and moves to get Case:
|(3)||a||Maggie1 might [vP t1 mend the lawn mower]|
|b||for Tony1 to [type the letter] (would be helpful)|
|c||Harry1 had [vP t1 helped the police]|
A phrasal position to the left of a head could be taken to be its specifier. Clearly the subject is a phrase and it always precedes the inflection at S-structure and hence we might assume that the position to which the subject moves, when it leaves the VP, is the specifier of the IP:
Note that apart from complementisers, which we will discuss in the next chapter, and adverbials, which we will discuss at the end of the present chapter, this structure accounts for all elements of the clause. Specifically we have a subject position, an inflection and a VP predicate: the three obligatory parts of the clause. It seems reasonable to claim therefore that the IP IS the clause. This point of view addresses an issue raised in chapter 3 concerning the exocentric nature of the clause. There we discussed reasons for not considering the subject or the VP as the heads of the sentence as they do not seem to have the right properties of a head. Traditionally therefore it has been assumed that clauses are headless.
However, the traditional assumption is challenged by the analysis in (4), where it is claimed that clauses most definitely do have heads. There is much evidence to support this. Firstly consider the relationship between the inflection and the clause. The inflections come in two basic types: finite and non-finite. The finite inflections consist of the modal auxiliaries and tense morphemes. The infinitival marker to is non-finite, but we also get clauses, traditionally called participles, in which the inflection on the highest verbal element is either ing or en (or one of its irregular versions):
|(5)||a||we are anxious [for Sam to succeed]|
|b||the crowd watched [the fire brigade rescuing the cat]|
|c||I saw [the cat rescued by the fire brigade]|
In the previous chapter, we analysed the morphemes in (5b) and (c) as light verbs heading a vP and so the status of the embedded ‘clauses’ in these examples is unclear at the moment: they may be clauses (i.e. IPs) or they may be simple vPs. We will not attempt to deal with this issue here, returning to it in a later chapter. But there are some similarities between these clauses and the infinitival clause in (5a) which are useful to consider. Traditionally, those clauses containing a finite inflection are called finite clauses and those containing a non-finite inflection are non-finite clauses. Thus the relationship between the inflection and the clauses has been long acknowledged. I suspect that the relative semantic unimportance of inflections and the lack of recognition of their syntactic importance have contributed to the fact that traditional grammars have failed to recognise them as heads.
It is important to realise that there are differences between clauses headed by finite inflections and those headed by non-finite inflections to see that inflections really do have a contribution to make to the clause. To start, clauses headed by a finite inflection can be main clauses and do not have to be embedded, though they may be:
|(6)||a||Will won’t stop the car|
|b||I suppose [Will won’t stop the car]|
|(7)||a||*Tim to stop the car||–||I want [Tim to stop the car]|
|b||*Tim stopping the car||–||I watched [Tim stopping the car]|
|c||*the car stolen||–||I saw [the car stolen]|
In embedded contexts, we see another difference between finite and non-finite clauses in that a finite clause can act as the complement of the complementiser that, while only infinitival clauses can act as the complement of the complementiser for:
|(8)||a||… that [Karen could cook the dinner]|
|b||… that [Karen cooked the dinner]|
|c||*… that [Karen to cook the dinner]|
|d||*… that [Karen cooking the dinner]|
|e||*… that [the dinner eaten]|
|(9)||a||… for [Tracy to teach English]|
|b||*… for [Tracy teaching English]|
|c||*… for [English taught]|
|d||*… for [Tracy can teach English]|
|e||*… for [Tracy taught English]|
These data not only suggest that there is a difference between finite and non-finite clauses, but also that the infinitive and the participles have a different status, perhaps indicating that while the infinitive has an IP status, the participles are really vPs. The main point is, however, that different clauses distribute differently and this correlates with which inflectional element they contain. All this adds up to the conclusion that the inflection does behave like a head in that it projects its properties to the whole construction and as we saw in chapter 2 it is heads that do this.
So far we have taken the rather simple (perhaps simplistic) view that the VP is the complement of the inflection because the VP follows it. Indeed, if we assume that the VP is the complement of the inflection, this is exactly what we would expect to find as in English all complements follow the head. So this assumption accounts for certain word order facts of English that without it would simply have to be stipulated. Exactly the same is true for the subject. If we assume that this is the specifier of the inflection we account for why the subject precedes both the inflection and the VP, as this is exactly the position in which we find English specifiers.
To see the advantage of this analysis, consider what happens if we do not assume that the inflectional element is the head of the clause. English is often described as an SVO language, based on a way of classifying languages in accordance with the ‘typical’ ordering of the major elements of the sentence (subject, verb and object). Without X-bar theory and the notions of head, complement and specifier, however, this is just a description of the facts which tells us nothing beyond what can already be observed. Assuming X-bar theory we have a way of accounting for word order patterns by using general statements about the relationships between elements in an X-bar structure and so this is a step in the right direction. However, if we do not assume that the inflection is a head, it is not easy to think of how we can use X-bar generalisations to account for the basic word order of English. This is especially so if we take the traditional view that sentences are exocentric and therefore stand outside of the set of facts that X-bar theory can account for. Only if we assume that sentences are endocentric can X-bar generalisations be used to account for word order facts concerning sentences.
Thus we seem to be inevitably drawn to the conclusion that sentences have heads and that the elements of the sentence are organised in terms of X-bar relationships to the head. The only question that remains is ‘what is the head of the clause?’ and there seem to be very few options available. The only two real contenders are the inflection and the verb and of these only the inflection really satisfies all the conditions with the minimal number of assumptions.
More supporting evidence for the head status of the inflection comes from its relationships to the other clausal elements. As a head we should expect the inflection to impose restrictions on its complement and specifier positions. Of course, we would not expect these to be based on Θ-roles as the inflection is a functional element and plays no role in Θ-role assignment. Instead we would expect these restrictions to be similar to those found within the DP discussed in chapter 3. Recall that determiners always take NP complements and no other phrase can appear in this position. The complement of an inflection is always a verbal phrase, be it vP or VP and again no other phrase can appear in this position. We can make this more precise if we use categorial features to describe the situation. The phrase that sits in the complement position of the inflection must be headed by an element with the categorial features [(–F), –N, +V], that is, by a non-functional verb including V and v. We can therefore suggest a very restrictive template for the lexical entries of all inflections:
|(10)||category:||[+F, –N, +V]|
|subcat:||[(–F), –N, +V]|
Inflections also impose restrictions on their subjects. Again these restrictions are not thematic in nature but similar to those imposed by determiners on their specifiers. Recall that only a certain kind of determiner allows a specifier: the possessive determiners. The possessive position is restricted to genitive elements, as shown below:
|(11)||a||[DP his Ø [NP car]]|
|b||*[DP he Ø [NP car]]|
|c||*[DP him Ø [NP car]]|
|(12)||a||… that he will hew the rock||–||*… that him will hew the rock|
|b||… that he hewed the rock||–||*… that him hewed the rock|
|c||… for him to hew the rock||–||*… for he to hew the rock|
|d||… him hewing the rock||–||*… he hewing the rock|
|e||… him hewn||–||*… he hewn|
As we can see in (12) the subject of the non-finite clause appears in the accusative. There is a further possibility with non-finite clauses which is not available with finite clauses and that is to have a missing subject:
|(13)||a||Peter prefers [- to be dressed]|
|b||Lucy likes [- being dressed]|
|c||the artist painted the model [- dressed]|
|d||*they think that [- dressed the model]|
|e||*they hope that [- will dress the model]|
We will discuss the nature of these restrictions in a later section. For now the important observation is that the inflection imposes these restrictions and hence is demonstrated to have head-like properties.
|(14)||a||Larry dislikes citrus fruits|
|b||we likeØ them|
The form of the inflection in (14) depends on properties of the subject. This phenomenon is known as agreement (see chapter 1). In English, agreement is very restricted, visible only in the case of the present tense morpheme and the present and past tense forms of be. We saw in chapter 4 that the possessive determiner also shows a similar pattern, having one form for pronominal possessors and another for non-pronominal possessors:
|(15)||a||[DP Carl ’s [NP car]]|
|b||[DP his Ø [NP car]]|
For inflections what determines the agreement form of the inflection is the person and number properties of the subject. With a third person and singular subject the inflection is realised as (e)s and with any other subject it has a null realisation:
|(16)||a||[IP Carl does [VP not have a car]]|
|b||[IP we doØ [VP not have a car]]|
If we take agreement to be a relationship established between a head (perhaps limited to functional ones) and its specifier, these observations again lead us to the conclusion that the inflection is a head.