Chapter 6 - Inflectional Phrases(40c) is of course grammatical, but only with special stress on did and used to assert
something that has previously been denied. Thus, it does not mean the same thing as I
left and in fact cannot be used to mean this.
These observations show that this zero morpheme is in complementary distribution
with tense and thus the straightforward conclusion is that it IS tense. But how can this
be if tense is an inflectional element and the zero morpheme is not in complementary
distribution with modal auxiliaries, which are also inflectional elements? What the
data show is that it is not modals that tense is in complementary distribution with, but
the zero tense morpheme that accompanies the modal and hence the conclusion is that
if modals are of the category ‘inflection’, then tense is not of this category. Given that
tense is situated in front of the VP, we can assume that it is a head that selects a verbal
complement and given that it follows the inflectional elements (i.e. modals) it must
project a verbal phrase. In other words, tense is yet another light verb:
(41) IP
DP I'
I vP
modal v'
v VP
tense DP V'
V
This analysis raises the question of what category ‘inflection’ is if it excludes the tense
morpheme, and specifically what occupies this position when there is no modal? To
answer this, consider the properties of modal auxiliaries. It is a traditional idea that
they are not actually in complementary distribution with tense, as in some sense they
display a kind of tense inflection:
(42) may might
can could
shall should
will would
(must)
Virtually all modals come in pairs, which might be claimed to represent a distinction
between past and present. The use of these forms supports this view:
(43) a I think I am going
b I thought I was going
c *I thought I am going