Ramzi Elkawa
September 13, 2016
Introduction The set of data considered in this paper concerns the typology of phrases
that can be embedded in sentences and the additional phrasal structure required to fit this
data neatly with TP domination and the rules of an X-bar schema.
By the end of this paper the theoretical groundwork will have been laid for a phrase
called a CP with a head called a C complementizer that allows the embedding of sentences. Along with that and if, the null head is a C as well and carries syntactic information
in a transformational grammar.
In English C and T heads share a special relationship. It isnt obvious that this should
be the case in all languages, but this paper will deal with the phenomena as it appears in
English alone.
Theoretical Grounding This paper will consider its data with the assumption of several
X-bar Phrase rules along with some assumptions about complements and subcategorisation.
(1)
a.
b.
c.
d.
TP > DP T0
T0 > T VP
VP > V0
V0 > V {VP, DP, PP}
These are the assumed PS rules. The internal structure of DPs and PPs will be ignored
as trivial to the cases discussed here. Importantly, whether a verb takes either of them,
neither or even both as complements is decided by a verb heads subcategorisation. This
is lexicalised information. T heads subcategorise in much the same way, but rather than
subcategorising for different phrase complements, they subcategorise for different forms of
their V heads. Crucially, subcategorisation only has power over complements within the
realms of this theory.
In part 1 we will only consider one hypothesis, as it easily accounts for the data. There
is something called a CP. Its internal structure will have a C head and a TP complement so
as not to have a problem with infinite generation as seems necessary in adjunction. We will
assume that it is a V head complement for reasons that will become clear in the analysis.
In Part 2 we will initially consider two hypotheses in order to fully explore the data.
1
(2)
a.
b.
Though we will decide on the second hypothesis, considering a) will allow us to address
why a transformational approach is necessary
Part 1 Our first hypothesis provides an analysis for how (3)s that structurally interacts
with the sentence. This structural proposition is realised in (4).
(3)
(4)
The power of this analysis though comes from the pattern of ungrammaticality we see in
(5) and (6).
(5)
(6)
Though sentences like (7) and (8) suggest that both whether and for pattern as Cheads, something about the verb insisted is blocking their use in this context. This
heavily suggests subcategorisation on the part of insisted and remaining within the realms
of our theory requires us to say that CP is a complement of insisted in order to account
for this subcategorisation.
(7)
(8)
The combination of (3) (5) and (6) then allow us to make a partial lexical entry for insist
as it subcategorises for Cs. We can combine this with more patterns of ungrammaticality
and create a table.
Verb
Insist
Manage
Wonder
Know
Like
Allowed Cs
that
for
whether
that, whether
for
ungrammatical sentence. This requires that for is selecting for to, this also necessitates
complement structure.
Part 2 Data in part 2 focuses on an interaction between embedded phrases and T-heads
showing up in C-head positions. (9) displays an example of this happening. (10) displays
that some verbs that do subcategorise for CP complements, do not allow T-heads to appear
in this C-position. (11) shows that the verbs that subcategorise for T-heads subcategorise
for if and whether too. (12) shows that C-positioned T-heads and C-heads cannot co-occur.
(9)
(10)
(11)
(12)
The most straightforward assumption about this data to take would be Hypothesis 2a as
displayed above. That is, T-heads can also appear as C-heads. They are subcategorised
for in much the same way (e.g. wonder allows T-type C-heads, think doesnt). The
semantic similarity between (9) and (11) is a coincidence. And, (12) is ungrammatical
because a C cannot take multiple heads. This hypothesis is represented in (13).
(13)
see tree 2
This analysis works fine then, unless we account for (14) and (15), which are ungrammatical, despite coherence with Hypothesis (2a)
(14)
(15)
(16)
see tree 3
(17)
see tree 4
This works well, except for one more problem. (18) provides a sample of what looks like
an empty C-head that post-transformation is ungrammatical (10).
(18)
Though this seems to be a problem for the transformation theory, the solution provides
insight into the syntactic difference between declarative clauses and polar questions. That
is there must be a bare C-head that prompts transformation [+interrogative] and one that
does not [+declarative]. So DS (18) and (9) really look like (19) and (20)
(19)
see tree 5
(20)
see tree 6
Since there are declarative C-heads then, that syntactically flag the semantics of a clause,
it seems plausible that every sentence really has the root node CP. Therefore the difference
between:
(21)
(22)
Takes place in the head of the root node of each DS. (21) may then look like (23), while
(22) DS looks like (24)
(23)
see tree 7
(24)
see tree 8
T-heads then come to occupy C-positions via transformation. This can be observed via
the remnants they leave behind in sentences, like bare verb forms. Normal V-heads do not
seem to undergo this transformation. The fact that auxiliary verbs do though creates a
question for another paper.
Conclusion CP is a robust phrase type that allows for embedded sentences. It also
confirms our intuitions that the declarative and polar question forms of a single declarative
statement are deeply syntactically similar. This phrase follows X schema and has many
different heads. These heads subcategorise as well. Thinking with this CP also opens the
door to start thinking about transformations generally.
Tree
1
Tree
3
Tree
5
Tree
7