Document Sample
Subjects Powered By Docstoc
					Non-agentive subjects
1     What we know about subjects so far

As we have seen, the subject position is occupied by some argument moving to it and so it is
vacant at D-structure, before movement takes place. The argument that moves originates
inside the verb phrase, either in specifier of VP or vP.

The reason why an argument moves to the subject position is that it lacks Case in its original
position and the subject position is where it can get Case: nominative from the finite I and
exceptional Case in infinitival clauses. Hence, under the assumption that it is the light verb
that assigns Case, it will be the argument in the spec of vP that will move when the light verb
is present as this will be the Caseless argument:

(1)          vP

        DP        v’

             v         VP

                  DP        V’


In the absence of the light verb, the argument in spec VP will move as it will then lack Case:

(2)          VP

        DP        V’


So far we have concentrated on agent and theme arguments. The agent is introduced by the
agentive light verb and hence will be subject when present and the theme is generated in the
specifier of the VP and hence will be subject in the absence of a light verb and object when
one is present. However, other elements can be subjects and objects and this is what we will
be looking at here.

2     Non-agent Subjects

We start with the subjects, as some of the observation we will make concerning these will
provide possible analyses for double object constructions.

2.1    Non-argument subjects

Not all subjects are arguments. Sometimes a subject can be a meaningless place filler, known
as a pleonastic or expletive subject:
(3)    a          it is snowing
       b          it seems that John is rich
       c          there is a man at the door

A number of questions arise concerning these kinds of sentence. The obvious one is why
these meaningless subjects are there in the first place, given that they add nothing to the
semantic content of the sentence. Obviously their function is grammatical rather than
semantic and it seems that the grammatical principle that they fulfil is the requirement that
there be a subject. It seems to be a requirement of English that all clauses have subjects, i.e.
that the specifier of IP is filled by something, which is standardly referred to as the Extended
Projection Principle (EPP). Typically this requirement is satisfied by the Caseless argument
moving to this position, but in (3) it seems that no argument can undergo this process. Thus
the EPP must be satisfied in another way and this is the function of the pleonastic element.

But this gives rise to another, more basic question: why can no argument move to subject in
these cases? There is no single answer to this question and the sentences in (3) represent three
different cases. Let’s start with (3a). This concerns a special verb (to snow) of a subcategory
that appears to take no argument at all. Such verbs tend to refer to the weather or atmospheric

(4)    it rained/thundered/hailed/stormed/...

Obviously, if these verbs have no arguments, no argument can move to the subject position
and hence we would expect the pleonastic subject to appear.

Yet this account might be a little too simplistic. The idea that there are predicates which take
no arguments is more than a little disturbing. Standardly, predicates are said to name states or
relationships that hold of their arguments. A predicate with one argument (a one place
predicate) may names a state or property of that argument or an action performed by it:

(5)    a          he is tall
       b          he grew
       c          he sneezed

A two place predicate may refer to a relationship which connects the two arguments:

(6)    a          John is next to Mary
       b          John knows Mary

What could a zero place predicate possibly mean? Moreover, this ‘pleonastic’ subject can
appear with predicates which do take arguments. The following is ambiguous depending on
whether it is the weather that is being talked about or a specific object:

(7)    it’s hot

If this was used to warn someone not to touch a saucepan on a stove, for example, then the
pronoun subject would be referential, referring to the saucepan. But if this was a statement
about the weather, the pronoun seems to have the same status as the subject of a weather
verb. But there are not two predicates in use here, one which takes and argument and one
which does not. So the question is how can a meaningless subject appear with a predicate
which assigns a thematic role? Finally there is evidence that the subject of weather predicates
does not behave like other ‘meaningless’ subjects in that they are able to control a
phonologically empty pronoun subject of an infinitival clause. Compare the following:

(8)    a       they sank the ship [ PRO to collect the insurance]
       b       he (only) seems to be rich [ PRO to impress the neighbours]
       c       * it seems he is rich [ PRO to impress the neighbours]
       d       it rains [ PRO to water the plants]

In (8a) the PRO subject of the infinitival clause is controlled by the subject of the main
clause: it is the ones who sank the ship who will collect the insurance. (8b) involves a case of
raising and the raised subject he acts as the controller of PRO. (8c) is similar but without the
subject raising. The sentence is ungrammatical with the intended meaning similar to that of
(8b) demonstrating that pleonastic subjects cannot control PRO. However, the grammaticality
of (8d) is unexpected if this involves a pleonastic subject. We conclude therefore that the
subject of weather verbs is not really meaningless.

The subjects of weather verbs have been termed ‘quasi arguments’ as they do not seem to be
fully referential arguments but they are not meaningless either.

Returning to the examples in (3), and as is shown in (8c), some uses of it as a subject really
are pleonastic. The examples given involve raising verbs in which raising does not take place.
Exactly the same can be seen in certain passives which do not involve movement:

(9)    a       We believed [that John was rich]
       b       John was believed [ t to be rich]
       c       it was believed [that John was rich]

As we know, when a verb is passivised the light verb associated with the external argument
goes missing. In cases like (9) this can trigger the movement of the subject of the
complement clause, if this is infinitival, or the insertion of an it when the complement clause
is finite. Obviously as there is no external argument in these cases, the it subject is not at all
semantically related to the passive verb and so it really is pleonastic.

We have accounted for the difference between (9b) and (c) in terms of the Case of the
subject: in the infinitival clause, there is no Case assigned and the subject must raise, thereby
filling the subject position of the main clause and satisfying the EPP. When the complement
clause is finite, its subject receives Case from the finite inflection and therefore does not need
to move. In this case the EPP must be satisfied by the insertion of the pleonastic subject.

But again, all is not quite as simple as this. Although there might be a reason why the subject
of the finite clause does not move, it is not clear why there is no movement at all in these
constructions. To see this, compare the following:
(10)   a       it seems [that John is rich]
       b       * [that John is rich]1 seems t1

(11)   a       it helps [that John is rich]
       b       [that John is rich] helps

Obviously in the case of raising, it is impossible for the clausal complement to move to
subject position to satisfy the EPP. The data in (11) however seems to indicate that such
movement is possible in other cases.

An old account of this difference claimed that while the clausal argument of the raising verb
originates in complement position, that of others originates in subject position. From this
point of view, it is the structure with the pleonastic subject that is derived by movement of the
clausal subject to some position behind the verb known as extraposition:

(12)                 IP                                           IP                CP

              CP            I’                               t1        I’    that John is rich1

       that John is rich helps                        it          helps

Here the first structure indicates the base position of the embedded clause and the second its
movement behind the verb (we will not be concerned with the question of where the clausal
moves) with the pleonastic subject inserted after the movement takes place. However, under
present assumptions, this cannot be maintained. To start we are assuming that the subject
position is underlyingly empty and therefore no argument originates there. Moreover, current
assumptions about traces should mean that not only would the insertion of the it subject be
superfluous, as traces satisfy the EPP when they are in subject position, but it should actually
be impossible to insert the pleonastic in to a position already occupied.

Perhaps the best way to translate this analysis into the current system would be to claim that
the clausal argument of a verb like help is an external argument, and therefore introduced by
some additional light verb, whereas the argument of a raising verb like seem is an internal
argument originating somewhere in the VP:

(13)                vP                                                       VP

              CP            v’                                          CP           V’

       that John is rich v VP                                that John is rich       V

                          e helps                                                  seems

Of course it still remains to be explained why the clausal argument can move from the
specifier of vP but not from that of VP, something we will not attempt here.

The last kind of meaningless subject exemplified in (3) concerns the pleonastic use of there.
As the syntax of there-constructions is rather involved, we will deal with it separately in the
following section.
2.2    There constructions

We have noted that the ability to appear in a there construction is a diagnostic of
unaccusative verbs, as intransitives do not appear in such constructions:

(14)    a      there arrived a letter
        b      * there sneezed a man

The observation seems to suggest that as the argument of an unaccusative is a theme, it can
surface in the object position, which is not a possibility with the agent argument of the
intransitive. However, it is not at all clear why this difference should hold when one
considers the facts more carefully.

Our analysis of the intransitive has the agent moving to the subject position because it is not
Case marked in its original position. Therefore we claim that the reason for the
ungrammaticality of (14b) is that the agent lacks a Case and therefore violates the Case Filter.
As (14a) is grammatical, it must be the case that the theme argument is Case marked in this
construction, but as unaccusatives lack a light verb, it is not at all clear where this Case
comes from. Furthermore, if we propose a solution to this problem we have to account for
why the same solution is not available for agent of the intransitive verb.

The theme’s post verbal position indicates that the verb moves to some position in front of it.
Again, though, in the absence of a light verb, it is not clear where the verb moves to. Note
that we cannot account for the verb’s position with the assumption that it moves to I, as it still
precedes the theme when the I is independently filled:

(15)    there will arrive a letter on Tuesday

There is another observation concerning there constructions that we need to account for. The
post verbal theme is restricted to indefinites:

(16)    a      there appeared a light in the distance
        b      * there appeared the light in the distance
        c      the light appeared in the distance

Apart from the problem of accounting for this fact, this phenomenon raises a practical
problem. In order to provide an account of where the post verbal theme gets its Case from, it
would be helpful to know what Case it gets. Does it get the nominative Case of a subject or
the accusative Case of an object? The problem is that the facts conspire to make this an
unanswerable question: English overtly realises Case only on its personal pronouns; personal
pronouns are definite; only indefinite DPs can appear post verbally in there constructions.
Hence the vital data is unavailable with both of the following possibilities ruled out for
reasons nothing to do with Case:

(17)    a      * there arrived him
        b)     * there arrived he
Finally there is the question of the pleonastic subject used in these constructions: why is this
there as opposed to it, which is used in other cases where the subject position is unfilled by
an argument. Clearly the construction doesn’t work if we insert it into the subject position
and moreover, there cannot be used in other constructions involving unfilled subject

(18)   a       * it arrived a letter
       b       * there seems that John is rich

One attempt to account for the there construction claimed that what makes there different
from it is an ability to transmit its Case to another element in the sentence. Thus in the there
construction, the pleonastic subject sits in a nominative position and transmits this to the post
verbal theme:

(19)   there may arrive a stranger

This predicts that the post verbal theme has nominative Case, though we have no way of
checking this. It accounts for the grammaticality of the post verbal argument and also for why
there is used and not it. However it says nothing about the indefiniteness of the theme and
moreover it suffers a number of drawbacks which make it an unlikely solution to the
problem. First it is rather stipulative, proposing a mechanism whose sole purpose is to
account for this particular construction. Once it has been proposed moreover, it then has to be
explained why it is restricted to this construction only: if there can transmit Case to Caseless
arguments, why do we not see it in passive and raising structures which also have arguments
originating in Caseless postions:

(20)   a       * there was seen John                  (John was seen)
       b       * there seems [John to be rich]        (John seems to be rich)

Indeed, there is no reason why it shouldn’t be available for all structures allowing all
arguments to remain inside the verb phrase, even the agent:

(21)   * there smiled a man / there may a man smile

Furthermore, if the pleonastic subject allows the theme to remain in its original position, we
would expect it to precede the verb in cases where the verb does not move out of the VP:

(22)   * there will [VP a letter arrive]

All in all then, this analysis leaves much to be accounted for.

In an attempt to account for the indefinite nature of the post verbal theme, Belletti (1988)
claimed that there is a special Case, partitive, assigned to this argument. She states that this
Case is incompatible with definite DPs for semantic reasons and points out that in languages
which mark it overtly, it only occurs on indefinites. While this offers an understanding of this
restriction, it says very little about the rest of the construction: why, for example, does the
construction have a there subject and why can partitive Case not be assigned to other
arguments in the verb phrase.

So, is there a solution which can solve all these problems? Let us start with the observation
which the other theories have not even touched on: why is the there construction available for
unaccusatives but not intransitives? The difference between these is the presence of the
agentive light verb in the intransitive, thus it seems that the there construction is in
complementary distribution with the agentive light verb. In turn, this suggests that at the heart
of the analysis of the there construction is a light verb whose presence is blocked by the
agentive. This is a little like the analysis of the passive, where the passive morpheme is in
complementary distribution with the agentive light verb. Indeed, as we have seen, the there
construction is incompatible with the passive also. All this suggests the following analysis:

(23)       vP                                       vP                              vP

       DP       v’                                  v’                            DP       v’

            v        VP                         v        VP                      there v        VP

            e DP V’                            -en DP V’                               e DP          V’

                      V                                       V                                      V

The first structure represents a transitive verb with an agentive light verb. In this the light
verb assigns Case to the theme and the agent moves to subject position to get nominative
Case. The second structure is that of the passive, where as the passive morpheme/light verb
does not assign a theta role it does not assign Case either. Therefore the theme in the specifier
of VP gets not Case and has to move. The third structure is the there construction. As the
theme gets Case in its base generated position, we assume that this comes from the light verb,
as it does with the transitive. However, the light verb here differs from the agentive as it is
not thematic. Unlike the passive, however, it is associated with something in its specifier
position: the there. In a sense, there is like a quasi argument of this light verb: something
which is selected for but which doesn’t have referential content. Evidence that this is so
comes from the fact that there can licence a PRO subject in much the same way that quasi
arguments can but pleonastics cannot:

(24)   a        it rains [ PRO to water the plants]
       b        * it seems that John is rich [PRO to impress the neighbours]
       c        there lies a dragon in the cave [ PRO to protect the treasure]

Suppose that this is enough to allow the light verb to assign Case, but not enough to allow it
to assign accusative. If we assume therefore that the light verb assigns partitive Case we can
further account for the indefiniteness of the theme along the lines of Belletti. Finally,
assuming that the light verb involved here is like other abstract light verbs in English, the
unaccusative verb will have to move to it to support it. We therefore account for the order of
the verb and the theme under these assumptions.
The full analysis of the there construction we arrive at is as follows:

(25)            IP


                I             vP

               may DP               v’

                     there v             VP

                              -e    DP         V’

                                   a letter V


The theme receives partitive case from the light verb which selects the there quasi argument.
The theme therefore does not need to move, but is restricted to indefiniteness. The main verb
will move to support the light verb and hence will ultimately precede the theme. The quasi
argument is a DP and therefore needs Case and so it will move to the subject position to be
assigned nominative by the finite inflection.

2.3    Locative Inversion

Another construction used to identify unaccusatives is the locative inversion structure in
which the locative argument of the verb appears to be in subject position and the theme
remains inside the VP:

(26)    in the woods lived an ugly ogre

If we assume that the locative argument starts off as complement to the main verb, then it
seems that what happens is as follows:

(27)                 IP


                          I              VP

                      may DP                     V’

                              an ogre V                PP

                                          live in the woods

But as we can see, there are some problems with this analysis. Most obviously this predicts
that the verb should stay behind the theme in cases where the inflection position is filled. But
this is not the case:
(28)   a         in the woods may live an ogre
       b         * in the woods may an ogre live

Moreover, if the theme does not move to subject position, how can it escape the Case Filter
as there is nothing to assign Case to it? Finally, why would the locative PP move to the
subject position given that PPs do not require Case?

There is another observation which is rather puzzling: it seems that the same indefinite
requirement holds of the theme in the locative inversion structure as it does of the there

(29)   a         the/an ogre lives in the woods
       b         in the woods lives an ogre
       c         * in the woods lives the ogre

We have said that this restriction is the result of partitive Case being assigned. But why
would partitive Case be assigned in this construction given that the element which assigns it
is the light verb associated with the there construction?

It turns out that locative inversion is not incompatible with the there construction:

(30)   in the woods there lives an ogre

If the locative inversion structure really is the there construction, all the problems are solved:
the PP does not move to subject position, the theme gets partitive Case from the light verb,
the main verb moves in front of the theme to support the light verb:

(31)        PP              IP

       in the woods1               I’

                              I         vP

                            may DP                v’

                                  (there) v            VP

                                          e    DP           V’

                                              an ogre V          PP

                                                        live     t1

Thus, while the locative PP moves to some position in front of the subject (the exact position
of which we will discuss in a future lecture) the subject position is actually filled by the
pleonastic element. Apparently this does not have to be overt when the locative argument
undergoes fronting. While this is something that cries out for explanation, I will not attempt
to do so here. I will merely point out that null subjects in finite clauses are not unheard of in
many languages. For example Hungarian often has null subjects in finite clauses:

(32)    (én) nem értem

Generally, languages either allow such null subjects or they do not. Those that do (most
Romance languages, Slavic languages, Semitic languages, etc.) are known as ‘pro-drop’
languages, as they allow the pronoun subject of finite clauses to be dropped. English is not
normally pro-drop as overt subjects are required in most of its sentences. German is also not
normally considered a pro-drop language, but there are instances where a pronoun subject can
be dropped grammatically:

(33)    Gestern wurde lange diskutiert.
        yesterday was-3sg long discussed
        yesterday it was discussed at length

Specifically it is the pleonastic subject that is dropped here. So while German cannot drop
meaningful pronouns, like most pro-drop languages can, it is able to drop meaningless ones.
The same seems to be true of English, though in a far more restricted set of circumstances:
English can drop the there subject when it is preceded by a fronted locative argument.

2.4    Experiencer subjects and multiple light verbs

For the majority of the time we have so far concentrated on agent, theme and meaningless
subjects. But there are other arguments which can appear in the subject position. One of the
most common is the experiencer:

(34)    a      John felt the pain
        b      John saw Mary
        c      John believes in magic

In many ways, the experiencer behaves much like the agent: when there is an experiencer and
a theme, it will be the experiencer that is the subject, not the theme:

(35)    a      John fears commitment                   (John ate the sandwich)
        b      * commitment fears John                 (* the sandwich ate John)

Moreover, verbs with experiencer subjects can passivise and when they do, it is the
experiencer that goes missing:

(36)    a      John saw Mary                           (John kissed Mary)
        b      Mary was seen                           (Mary was kissed)

All this argues that we should treat the experiencer in exactly the same way as the agent: that
it is introduced by a light verb which takes the VP as its complement:
(37)           vP

        DP          v’

        John v           VP

                e    DP       V’

                    Mary V


We might suppose then that the only difference between the agent and the experiencer is
simply interpretational: experiencer is how we interpret the agent of certain verbs.

Yet, this is challenged by the fact that experiencers and agents are not in complementary
distribution: there are some verbs which have both agent (or causer) and experiencer

(38)    the clown frightened the children

In this case, the clown is interpreted as the one performing an action (agent) and the children
are the ones who experience and emotional reaction (fright). Note that the agent is the subject
and the experiencer emerges in object position and it would be ungrammatical to have it the
other way round (i.e. to interpret (38) as the children doing the frightening and the clown
being the one who is frightened).

But if we hold to the UTAH, we cannot analyse this as having the experiencer in the specifier
of VP (standard object position) as we have seen it appears in the specifier of a light verb in
other constructions.

The most obvious solution would be to suggest that in cases like (38) there are two light
verbs, one which introduces the agent and one for the experiencer:

(39)      vP

   DP               v’

the clown v               vP

           e         DP             v’

               the children v            VP

                               e frighten

Note that this analysis entails that the main verb has to move twice to support both light
verbs. This is not problematic as we are assuming that heads can move multiple times to
support other morphemes, such as tense and agreement. In this configuration, note that the
experiencer will receive Case from the agentive light verb, and hence we can account for why
it is the agent that has to move to the subject and not the experiencer. Obviously when there
is no agent, (as in the structure in (37)) the experiencer being the top most argument will lack
Case and therefore it will have to move to the subject position.

The issue that this analysis raises however is whether it is possible for there to be multiple
light verbs. So far the indications have been that there is only one position for a light verb and
hence light verbs of different natures have been in complementary distribution with each
other: the agentive and the passive light verbs cannot co-occur and the passive light verb
cannot co-occur with that of the there construction. One would think therefore that a structure
such as (39) should be impossible.

However, it should be noted that it is possible to have more than one overt light verb in any
one sentence:

(40)    John made Mary let Bill take a walk

In principle then the structure in (39) is a possible one and indeed what this indicates is that it
is the complementary distribution of agent, passive and there light verbs that needs
explaining, yet another issue we will not address.

2.5    Conclusion: what makes a subject?

We have seen in this lecture that the argument which moves to the subject position is always
the top one in the complex verb phrase. This follows straightforwardly from the assumption
that light verbs assign Case to the specifier of their complement. Thus, for every argument
except the top one, there will be a light verb to assign it Case and it will not have to move.
For the top argument there will be no light verb above it (otherwise it wouldn’t be the top
argument as that light verb would introduce a higher argument!) and hence it will not get

However, we have also noted that there is a definite pattern to which argument will be the top
one. The theme will only be the top argument when there is no other argument. This follows
from the assumption that it is the theme that is introduced by the main verb and the other
arguments are introduced by light verbs. However, there is also a definite order of arguments
introduced by light verbs: the agent is always higher than the experiencer. This seems to
reflect a hierarchy of arguments to the following pattern:

(41)    agent > experiencer > theme

The same hierarchy has been noted in other linguistic phenomena so it seems to have a
certain validity. What is uncertain is how the thematic hierarchy interacts with the principles
of structure to ensure the relevant ordering of arguments in any sentence.

Shared By: