12. Other Movements following Subjacency Condition
[1] Subject Raising: The type of NP-Movement called ‘Subject Raising’ in seem-clauses
or be likely-clauses can be analyzed appropriately by Subjacency Condition:
(86) The fur seems to be likely to fly
(87)
Given the classic idiom chunk argument as ‘The fur is flying’ or ‘The fur will fly’(means
‘A great commotion breaks out’), (86) will be derived from the underlying structure (87),
and the sentence (86) has three clauses: seems-clause, be likely-clause, and fly-clause.
As shown in underlying structure (87), the subject NP positions in seems- and be likely-
clause are empty, and so the subject ‘the fur’ in fly-clause can be moved into the empty
NP position. Specifically, let’s see the tree diagram.
The movement of the NP ‘the fur’ directly S1
into the empty NP main clause (S1) subject NP VP
position will involve an movement across the V S’
two bounding nodes S3 and S2, and therefore, seems COMP S2 bounding
nodes
it will violate Subjacency Condition. [-WH] NP VP
However, if the NP ‘the fur’ moves first into ② to V’
the empty NP position in the be likely-clause V S’
(crossing S3) on the S2 cycle, and then moves COMP S3
from that position into the empty NP position be likely [-WH] NP VP
in the seems-clause (crossing S2) on the S1 DET N to V’
cycle, then neither of these movements will ① the fur V
violate the Subjacency Condition, resulting in
(86) is well-formed. fly
♣ Why should the NP ‘the fur’ move?
Is it right that the NP should move only to fill the empty NP position? Consider the
reason.
[2] Relative Clause: Complex Noun Phrase Constraint for relative clauses or noun complement
clauses can be covered by Subjacency Condition.
The ungrammatical relative sentence (88) is derived from the underlying structure (89):
(88) *Who are you reading a book that criticizes ?
(89)
♣ There are two possible ways in which ‘who could be adjoined to the main clause COMP.
(1) First Possibility: The ‘who’, illustrated in ③, is adjoined directly to the main clause COMP,
but ‘who’ crosses three bounding nodes S2, NP1, and S1, violating Subjacency Condition.
(2) Second Possibility: ‘Who’ is adjoined firstly to the subordinate clause COMP on the S’2
cycle and adjoined secondly to the main clause COMP on the S’1 cycle, but this second
movement is across two bounding nodes, NP1 and S1, violating Subjacency Condition.
♣ These two possible analyses shows the same result, but the second analysis that follows the
successive cyclic movement is preferred.
[3] Sentential Subject: Subjacency Condition obviates the need for the Sentential Subject
Constraint(“No constituent can be moved out of a sentential sub-
ject””).
The Sentential Subject Constraint is originally designed to account for the ill-formedness
of the sentence like (90):
(90) *What would for me to do annoy you?
The underlying sentential subject ‘for me to do what’ have been regarded as an island.
So in accordance with the rule that any movement cannot extract an element out of the
island, the WH-Movement in (90) should not be permitted.
There are two ways in which ‘what’ might be adjoined to the main clause COMP1,
indicated schematically in the following structure: ③ is a direct movement whereas
① and ② is a successive movement.
(91)
The sentence (90) can be structured
as a tree diagram of (91).
(1) The First Way: ‘what’ is adjoined
directly to COMP1 as a single application of
WH-Movement. But the ‘what’ moves out
of three bounding nodes S2, NP1 and S1,
violating Subjacency Condition.
(2) The Second Way: ‘what’ is adjoined first to the subordinate clause COMP2(crossing
one
bounding node S2), and then to move from there to be adjoined to COMP1(crossing two
bounding nodes NP1 and S1). Here, the second adjunction violates Subjacency Condition.
♣ There is no way of deriving sentences like (90) without violating the Subjacency Condi-
tion,
which obviates the need for the Sentential Subject Constraint.
[4] WH-Island Structure: Another constraint which can be subsumed under Subjacency
Condition is the WH-Island Constraint in which no constituent can
be moved out of any clause containing a wh-phrase in COMP.
The sentence of (93) is ungrammatical under the WH-Island Constraint, since ‘what’
is moved out of a clause containing a wh-word ‘whether’ in COMP. The underlying
structure is shown in (94) with the type of movement:
(93) *What might he ask whether I hid ?
(94)
♣ There are two ways of the WH-Movement in (93):
(1) The First way: The ‘what’ is adjoined directly to the main clause COMP as in ③, but
this
movement crosses two bounding nodes S2 and S1, violating Subjacency Condition.
(2) The Second Way: The ‘what’ is adjoined first to the subordinate clause COMP as in ①,
and then to the main clause COMP as in ②; the first movement crosses S2, and the
second movement crosses S1. Only one bounding node is being crossed in each
movement. However, it turns out that (93) is ill-formed even though (93) doesn’t
violate Subjacency Condition.
Here, the solution is that the adjunction of ‘what’ to the subordinate clause COMP
‘whether’ is blocked by WH-Complementiser Constraint of (95):
(95)
♣ The constraint (95) blocks ‘what’ from being adjoined to the COMP ‘whether’; ‘what’
cannot be adjoined successive cyclically, and directly attached to the main clause COMP.
This adjunction violates Subjacency Condition, crossing two bounding nodes S1 and S2.
There is some independent support for WH-Complementiser Constraint (95). In (96),
we can adjoin either of the wh-word ‘who’ and ‘what’ to the main clause COMP indi-
vidually.
However, we cannot adjoin them both to COMP, since a constraint such as (95) would
provide a natural explanation for (96), in which (96) (d)~(e) violate the constraint (95):
(96) (a) is an underlying structure of (96) (a)~(e). If so,
What is the difference between the movement processes in (d) and (e)?
=> (96) (d)와 (e)는 Subjacency Condition과 무관하며 WH-Complementiser Constraint에 의해서만
문장의 비문법성이 설명될 수 있다
.
◈
Conclusion
It seems that the Subjacency Condition has considerable generality,
since it obviates
the need for a number of earlier constraints. Essentially, what the condi-
tion says is:
No constituent can be moved out of more than one containing bound-
ing node
S or NP in any single rule-application.
