Beruflich Dokumente
Kultur Dokumente
STANDARD ENGLISH
OXFORD STUDIES IN COMPARATIVE SYNTAX
Richard Kayne, General Editor
ALISON HENRY
246897531
Printed in the United States of America
on acid-free paper
For Mark Henry
This page intentionally left blank
Preface
1. Introduction 3
2. Subject-Verb Agreement 16
3. Overt-Subject Imperatives 45
4. For-To Infinitives 81
7. Conclusion 136
Notes 759
References 142
Index 147
This page intentionally left blank
BELFAST ENGLISH AND
STANDARD ENGLISH
This page intentionally left blank
1
Introduction
3
4 BELFAST ENGLISH AND STANDARD ENGLISH
and certain finite subordinate clauses occur with null subjects, in what is
otherwise a non-pro-drop language.
(9) They were lucky got away.
(10) You were as well took the job when you were offered it.
The analysis of these constructions is relevant to a number of current
issues within syntactic theory in general and English syntax in particular.
A number of characteristics of English which have been claimed to
derive either from universal principles or from parameter settings hold-
ing for English seem to differ in Belfast English. Thus, it has been
6 BELFAST ENGLISH AND STANDARD ENGLISH
claimed (by Rizzi & Roberts 1989 and Vikner 1991, among others) that
the unavailability of inversion in embedded questions in English derives
from the Wh-criterion (Rizzi 1991; for a discussion of this feature, see
Chapter 5). Although the Wh-criterion seems otherwise to apply in
Belfast English just as in standard English, inversion is freely available
in embedded questions in Belfast English, calling into question this
analysis unless some other factor can be brought in to account for its
availability. Similarly, a number of accounts have been proposed for
why main verbs in English (except be and have) never raise out out of
VP: for example, Pollock (1989) suggests that they cannot assign their
0-roles from higher positions because AGR is weak in English; but as
shown in Chapter 3, Belfast English imperatives seem to show such
movement, and any account of why verb raising is not generally possi-
ble must therefore take account of the fact that it does happen in this
case. Examining such structures in detail will allow us to see whether
the analyses proposed for standard English are correct, but additional
factors in Belfast English mean that the facts look different, or whether
in fact the standard English analyses do not hold up when faced with
data from another dialect.
Before we go on to look in detail at the differences between Belfast
English and standard English (and within Belfast English), it will be
useful to set the scene by considering some background issues. Thus,
although the study is concerned with the grammars of contemporary
speakers of Belfast English rather than the historical development of
that variety, it will be useful to consider briefly the historical and geo-
graphical background of this variety. This is done in the next section.
We also need to outline briefly the theoretical framework in which the
study is undertaken; this will take up a later section, titled "Principles
and Parameters Theory." The final section of this chapter considers the
special methodological issues and problems which arise in working on
non-standard dialects.
Belfast English is the variety spoken in and around Belfast, the capital
of Northern Ireland. Belfast is a major commercial and industrial centre
with a population of some half a million people.
English speakers in Belfast are largely monolingual, and there is no
Introduction 1
For many years I have been disgusted with the bad grammar used by
school-leavers and teachers too sometimes, but recently on the lunch-
time news, when a secretary, who had just started work with a firm, was
interviewed her first words were: "I looked up and seen two men" etc.
It's unbelievable to think, with so many people out of work, that she
could get such a job. ("Have went," Saintfield, N. Ireland)
Form; the differences between languages result from the fact that some
processes apply earlier in some languages than others; those processes
which apply early are overtly visible, while those which apply later are
not.
While earlier models envisaged words being introduced from the lex-
icon in their base form, and moving in the course of the derivation to
acquire Case-marking or to pick up affixes, the Minimalist model envis-
ages words being inserted from the lexicon in their fully specified,
inflected form, and moving for checking to affixes; once checked, the
affixes delete. What determines whether movement for checking
applies early or late is whether the affix in question is weak or strong. If
strong, it is visible at PF if undeleted, and it will be identified as an
unattached affix at that level, causing the derivation to crash; therefore,
where a functional element is strong, it triggers overt movement, before
what is called spell-out. If it is weak, movement does not occur until
LF.
The reason that movement is delayed until LF if possible is the Pro-
crastinate Principle: movement occurs as late as possible, LF movement
being in some sense "less costly" than overt movement and therefore
preferred.
Another important principle with a "least effort" flavour within Mini-
malism is that movement only occurs if forced; there is no optional
movement.
An important difference between the Minimalist approach and other
frameworks which will be relevant for our study here relates to Case.
Earlier approaches incorporated a Case filter operating at surface struc-
ture, which excluded structures in which an overt Noun phrase had not
Case. At S-structure, in order to have Case, NPs had to be in a Case
position, or be part of a chain of which one element was in such a posi-
tion.
While the Minimalist approach requires Case to be checked, this can
be done either in the overt syntax or later. Since NPs are inserted from
the lexicon with all their inflectional properties, the fact that an NP is
overtly Case-marked does not mean it has to be in a Case position or
part of a Case-marked chain. It can raise to check Case at LF, subjects
raising to SPEC/AGRS and objects to SPEC/AGR0. As we shall see, this
seems to work well for Belfast English where in imperatives, subjects
can be in a non-Case-marked position at S-structure, something which
would have presented a problem in earlier approaches to syntax.
Introduction 11
16
Subject- Verb Agreement 17
general possible both with raising verbs and with verbs which remain in
VP. However, it is impossible with raising verbs if an adverb intervenes
between the subject and verb; whereas (18a) is grammatical, as in stan-
dard English, (18b) is not.
(18) a. The children really are late.
b. *The children really is late.
Thus, it seems to be the case that the adverb position between the sub-
ject and the topmost projection of INFL which exists in English is
unavailable in singular concord.
There is not a general adjacency requirement between subject and
verb in singular concord, however; an intervening adverb is fine with
verbs which remain in the VP.
(19) The children really likes pizza.
(20) These books probably costs a lot.
It might be thought that singular concord is the converse of this; that is,
20 BELFAST ENGLISH AND STANDARD ENGLISH
That these are indeed first person plural subjects is seen from their cor-
responding reflexive forms.
This again indicates that the subject is not in the same position as in
non-singular concord cases.
Note that the unavailability of this adverb position in singular con-
cord is in fact an argument in favour of the existence of such a position;
Baker (1991) argues that there is a single adverb position and that in
sentences where apparently an adverb intervenes between a subject and
raised verb, this is because in fact the verb has not raised. Thus for
Baker the structure of (59) and (60) are identical to that of (61), where
the verb has not raised; but note that in Belfast English, the adverb posi-
tion between the subject and verbs which remain in VP is available in
singular concord structures, indicating that a different structure is
involved.
(61) a. The girls probably like coffee.
b. The girls probably likes coffee.
Thus, we have noted that in singular concord, the subject cannot have
overt nominative Case and the verb is not marked for agreement.
Assuming that the subject originates in VP, it seems that the subject has
not risen as high as SPEC/AGRSP; the question then arises of where the
subject is at spell-out. Apart from the difference just noted, raising
verbs (in the sense of Pollock [1989]) manifest exactly the same pattern
of placement relative to negatives and adverbs with singular concord as
they do in sentences with agreement, indicating that the verb has raised
outofVP.
(62) The eggs is not cracked.
(63) The eggs is probably cracked.
Chomsky (1992) argues that the verbs be and have raise for checking
in the syntax, because they are too weak semantically to be visible for
movement at LF. Since verbs are marked for tense in the singular con-
cord construction, be must be at least as high as the Tense position in
the overt syntax; otherwise, since movement is unavailable at LF, the
V-feature of tense would remain unchecked at LF, and the derivation
would crash. Given the subject-verb order, the subject must occupy a
node higher than Tense. The obvious candidate here is SPEC/Tense P.
Subject- Verb Agreement 27
and that this subject trace is Case-marked. There is at first sight a prob-
lem with allowing Case-marked positions in NP chains if one is to
maintain the least-effort principle: it may not be clear why an NP would
raise if its Case has already been checked. It may be that the condition
should be re-formulated to refer to potential Case-marking positions
rather than actual traces. Alternatively, it may be that the requirement
that the head of an NP-chain be Case-marked is not strictly correct. An
NP may be able to check Case in one position (e.g., SPEC/Tense) and
then move to another position to check agreement, so that the highest
position of the NP-chain is not the one which is Case-marked in fact.
However, there is another way of interpreting this, without reference
to subject traces. Suppose that, as Duffield argues, the negative element
left-adjoins to TP at LF. Now in Belfast English, if no further raising
takes place at LF, then at LF the negative element will be in the follow-
ing configuration:
(72) [TPNeg[TP Any country [Tcould [VP stand it]]]]
If the subject has already checked Case, and the (non-agreeing) verb
does not require to check agreement, then there is nothing to force LF-
raising; the subject and verb can remain in Tense P, and presumably
Subject-Verb Agreement 29
AGRS will delete since it will not be required for semantic interpreta-
tion, leaving the LF representation as in (72).
One way to distinguish between these two analyses would be to con-
sider what happens when there is a plural NPI subject. If as we have
argued the subject in singular concord is in SPEC/Tense P, then it
should be impossible to have subject-verb agreement with an NPI sub-
ject. That is, a sentence like (73a) should be impossible, whereas the
corresponding sentence without agreement should be possible.
(73) a. *Any animals aren't coming.
b. Any animals isn't coming.
This seems to be the case; NPIs are only possible with non-agreeing
verbs, indicating that it is not the potential availability of Case-marking
in a certain position, but rather the actual presence of a Case-marked
element in the structure, that is the important issue.
If an approach along these lines is right, then it would be, as Duffield
points out, an important step towards reformulating the conditions on
NPI licensing as LF, rather than S-structure, conditions, for it will be
observed that at S-structure, the negative element is below the negative
polarity item; for example in the following sentences, any is above n 't
and not, which are presumably in Tense (with did) and in NEG, respec-
tively.
(74) a. Any student didn't apply for the job.
b. Any student did not apply for the job.
Under the Minimalist program, S-structure does not exist as a sepa-
rate level and all conditions are interface conditions, applying at LF or
PF. The availability of negative polarity items in subject position in
Belfast English seems to indicate that this is the case; at S-structure the
NPI in subject position is not c-commanded by the negative element, as
would be required by the S-structure condition on negative polarity
items. However, if the condition is an LF condition, and negatives raise
to adjoin to TP, then we have a natural way to account for the possibil-
ity of NPIs in subject position in standard English, and their impossibil-
ity in standard English. NPIs are possible in Belfast English because the
subject can be in SPEC/TP at LF.
That there is a link between singular concord and NPI-licensing in
subject position seems clear from the the fact that the two phenomena
seem to go together in speakers' grammars; those speakers who allow
30 BELFAST ENGLISH AND STANDARD ENGLISH
seen as removing the -s ending from forms other than the third person
singular. This goes some way to explaining the puzzle, pointed out by
Kayne (1989), of why in English the third person singular form, which
in other languages is often unmarked, is the one which has an overt
affix; in fact, the third person singular is the unmarked form, showing
only a tense inflection, with other forms being for agreement.
Notice also that our analysis, which has the subject in singular con-
SP, argues against the view that the
subject raises out of VP because agreement is obligatory in English
(Kitagawa 1986); in singular concord the subject raises out of VP oblig-
atorily even though it is not raising to SPEC/AGRSP and it is not trig-
gering agreement. It thus seems that what forces the subject to raise
must be the need to check Case; since in Belfast English Tense option-
ally checks Case, it need not raise further than SPEC/Tense.
Singular concord is available with do-support. Recent work suggests
that do-support is a language-particular rule which inserts do to bear
Tense and Agreement when LF verb raising is impossible and raising in
the syntax is also excluded (Chomsky 1989, 1992; Pollock 1989), as for
example with negative sentences containing verbs other than be, have,
and auxiliaries. Singular concord is possible with do-support, suggest-
ing that do may be inserted in Tense, and is not restricted to insertion in
the highest head of IP.
Note that here them is the distal demonstrative rather than the personal
pronoun; them does not act as a nominative pronoun.
(92) Them's no good, are they/*them?
(93) *John and Mary are good friends, aren't them?
is well known, simple personal pronouns must occur between verb and
particle, whereas other NPs can occur either in this position or after the
particle.
(99) a. I phoned him up.
b. *I phoned up him.
(100) a. I phoned the owner up.
b. I phoned up the owner.
Note that the class of pronouns which may appear after the particle is
identical to the class which may have singular concord; thus co-ordi-
nates and other pronouns which are part of a larger NP, demonstratives,
and -uns pronouns in Belfast English do not have to occur between verb
and particle.
(101) My friend phoned up him and her.
(102) They helped out us students.
(103) The staff tore up those.
(104) They phoned up yousuns.
argues that the NP is base-generated between the verb and the particle,
with the NP and particle forming a small clause construction; where the
NP appears after the particle at S-structure, this is the result of right-
ward movement, an operation normally reserved for "heavy" elements
and thus unavailable to weak pronouns. There are problems with the
analysis; as Johnson (1991) points out, the NP which occurs after the
particle here does not have to be as "heavy" as in other cases of heavy
NP shift.
This indicates that the verb does not have to be adjacent to the pronoun
at S-structure. We argue in the next chapter that these structures involve
the movement of the verb out of VP.
Moreover, we cannot require the subject to be adjacent to an overt
Case-assigner, or checker, at spell-out in either standard English or
Belfast English; for where the verb remains in the VP in the syntax, the
subject is clearly not adjacent to an overt Case-checker.
Koopman (1990) also has V-Prt-NP as the D-structure order, but
argues that the reason why pronouns must move in front of the particle
is that pronouns attract to SPEC positions—in this case to SPEC/PP;
note that if our analysis of singular concord is correct, then this cannot
be precisely the correct analysis, for in Belfast English the subject is in
a specifier position, SPEC/Tense, in the singular concord construction,
but this is not a possible position for the overtly case-marked set of pro-
nouns, which are forced to move to SPEC/AGRSP. We might amend
Koopman's analysis to require that pronouns attract to SPEC/Agree-
ment positions, but this would involve proposing that the position gov-
erned by a preposition in a Prepositional Phrase is a SPEC/Agreement
position, since these prepositions can occur there
(113) with him
(114) for them
As we shall show in the next chapter the verb has raised to C in these
constructions, and the object has moved to SPEC/AGR
subject cannot therefore be in SPEC/AGRSP; it appears to be in
SPEC/VP, but that is not a SPEC/AGR position.
Thus, a number of proposals have been made about the distinctive
placing of pronouns in verb-particle constructions, but none of these
precisely fits the Belfast English data. What we would need is an expla-
38 BELFAST ENGLISH AND STANDARD ENGLISH
nation for why a subset of pronouns must raise to AGRS, and appear
between the verb and particle in verb-particle constructions. Let us
begin by attempting to characterise the set of pronouns which must
raise to SPEC/AGRSP. The major difference between the elements
which must appear in the higher position and those which may also
appear in the lower position is that the former are for the most part
overtly Case-marked (or m-Case-marked, to borrow the term used by
Holmberg (1986) to distinguish morphological from abstract Case).
Thus demonstratives, the strong forms of pronouns which may occur in
coordinate structures, and -uns pronouns show no variation for Case;
whereas the personal pronouns which must appear in the higher posi-
tions are for the most part overtly Case-marked.
You, youse, and it are exceptions to the set, in that they do not have
overt Case-marking but nevertheless have to appear in the same posi-
tions as those which do; this suggests that, parallel to their paradigm set,
they require Case-marking which is in some sense overt but phonologi-
cally null. There is some evidence, however, that at least one member
of this set is changing to be a non-overtly Case-marked pronoun, and
that is Belfast English youse. For some speakers, this both permits sin-
gular concord and can occur after the particle in verb-particle construc-
tions.
(117) Youse is really stupid.
(118) I'll phone up youse.
The fact that weak pronouns are those which disallow singular con-
cord, and also must occur between verb and particle in the verb-particle
construction, and can object shift, suggests that it may be worth explor-
ing the possibility that what prevents the occurrence of nominative Case
with a singular concord verb is in fact a requirement that a pronoun sub-
ject raise to SPEC/AGRSP rather than a prohibition on nominative Case
as such. Let us consider, however, how we might distinguish between
the two possibilities.
The key to establishing whether it is nominative Case per se, or pro-
nouns, which are excluded with non-agreeing verbs seems to lie in
instances where pronouns are co-ordinated. We noted earlier that where
pronouns are co-ordinated, or where a pronoun occurs in a co-ordinate
structure with a full NP, the pronoun may either have default Case
(identical to the accusative) or bear the Case assigned to the whole NP.
studied by Kimball and Aissen (1971) and Kayne (1989). In this dialect
the verb may agree with a wh-element rather than with the subject.
(137) The people who Clark think are in the garden
Here we have a case where full NPs may lack agreement with the
,yerb, but nominative Case-marked pronouns have obligatory agree-
ment; in our terms, nominative pronouns must check Case in SPEC/
AGRSP, but full NPs, which are not overtly Case-marked, may check
Case in some other way.
Similar considerations apply to sentences in Belfast English and (col-
loquial) standard English where the subject appears to agree with
adjunct wh-words rather than the subject. As noted by Radzinski
(1985), sentences like the following may have singular verbs.
(140) Where's my glasses?
(141) How's the children?
It seems that in structures like (140) and (141) the verb is agreeing
with the wh-element; if this is again a case of agreement between the
wh-element and AGRS (which has moved to C), then the subject cannot
check Case in SPEC/AGRSP. Again, as we would predict, nominative
Case-marked pronouns are impossible here.
(143) a. *Where's they?
b. Where are they?
(144) a. *How's we doing?
b. How are we doing?
As with other alternations of this type, it is only nominative pronouns
which are required to be in the canonical Case-checking position.
42 BELFAST ENGLISH AND STANDARD ENGLISH
Conclusion
ject raises only as high as SPEC/Tense, where its Case can be checked.
However, nominative Case cannot be checked by Tense, and therefore
nominative Case-marked items cannot occur in this construction, but
must rather move to SPEC/AGRS.
The availability of SPEC/Tense as a subject position has two notable
effects in the grammar; it permits the occurrence of negative polarity
items in subject position, and the raising of objects where the verb
moves out of VP.
The impossibility of singular concord verbs moving to C shows that
head movement must be forced at each stage; a head cannot make an
unforced move to an intermediate projection just because that move is a
step which is necessary in order to check a higher projection.
3
Overt-Subject Imperatives
45
46 BELFAST ENGLISH AND STANDARD ENGLISH
Both of these facts make an analysis where the verb raises to C and
the subject to SPEC/AGR,, exactly as in questions but with a wider
range of verbs, less than straightforward: if this were the case, we
would not expect to find an object between the verb and the subject, and
we would always expect to find the subject after the auxiliary rather
than the lexical verb.
Before we consider what is happening in these imperatives, however,
we need to look at the facts in more detail, for Belfast English impera-
tives have not previously been discussed in the literature, and even stan-
dard English imperatives have not been extensively considered. We
thus proceed to outline the characteristics of overt-subject imperatives
in standard English and Belfast English, before going on to analyse
these in more detail.
Where an overt subject is used, it is most often you, but a small number
of other subjects are permitted, for example everyone, somebody, all of
you (Beukema & Coopmans 1989).
(24) Everyone put on their coats.
(25) Somebody tell me the answer.
(26) All of you keep quiet.
Vocatives often occur with imperatives, as in (27) and (28), charac-
terised by an intonation break between the vocative and the rest of the
imperative.
(27) a. John, put on your coat.
b. Put on your coat, John.
(28) a. You, tell me the answer.
b. Tell me the answer, you.
person singular, and in vocatives, this null element occurs in the true
subject position; being second person, it cannot be co-referential with a
third-person element. Similar considerations apply to reflexives. Where
there is a vocative third-person NP, a third-person reflexive is ungram-
matical; on the other hand, where a third-person NP is in subject posi-
tion, a third-person reflexive is possible.
(33) *Everybody, wash themselves.
(34) Everybody wash themselves.
Thus, it is clear that imperatives with true subjects are possible. And
as noted above, in standard English these always occur before the verb.
Negative imperatives have a number of interesting characteristics.
They always require do-support
(35) Don't do that.
(36) Don't go away.
Here, an overt subject, if present, occurs after don't
Note that where the imperative has an overt subject, the negative
must be the contracted form don't, and not the uncontracted do not
(43) a. Don't you speak to me like that.
b. *Do you not speak to me like that.
c. *Do not you speak to me like that.
Dialect A
Speakers of dialect A have a more restricted use of inversion than those
who speak dialect B; whereas dialect B speakers permit inversion with
any verb, in dialect A, inversion is only possible with a restricted range
of verbs. The verbs which permit inversion in this dialect are all intran-
sitive.
(57) Go you there.
is grammatical but
(58) *Read you that book.
with a transitive verb is not. Not all intransitive verbs permit inversion,
however, but only a subset of these. All of the verbs which permit
inversion in dialect A are verbs of motion; other verbs never permit
inversion in this dialect.
(59) *Eat you up.
(60) * Always laugh you at his jokes.
Not all motion verbs/verb phrases permit inversion, but only a subset of
these. It is at first sight difficult to characterise exactly which verbs per-
mit inversion, particularly as the same verb may or may not permit
inversion depending on the content of the remainder of the verb phrase.
For example, consider the verb run. In dialect A, this verb does not nor-
mally allow imperative inversion when it is the sole constituent of the
verb phrase.
(61) *Run you.
But it does allow inversion where it is followed by a particle such as
away, or an indication of destination.
(62) Run you away.
(63) Run you home.
(64) Run you into the garden.
It might seem that the generalisation is that, for some reason, at least
one constituent must appear after the subject. But in fact the presence of
Overt-Subject Imperatives 53
It seems that the type of constituent which follows the verb is signifi-
cant.
These facts are not exclusive to run; a similar pattern emerges with
most other motion verbs. However, there are a few verbs which allow
inversion regardless of the content of the verb phrase, for example,
arrive and leave
(67) Leave you now.
(68) *Run you now.
(69) Arrive you before 6 o'clock.
(70) *Run you before 6 o'clock.
The verbs or verb phrases which allow inversion have a semantic char-
acteristic in common. They all involve actions that have an end-point,
that is, they are telic (for a discussion of telicity, see van Hout et al.,
(1993).
This telicity need not be inherent in the verb, but it must be a charac-
teristic of the verb phrase of which it is part. This is why the presence of
an indication of destination, or of a particle which indicates that the
action has an end-point, is significant; these elements make the verb
phrase telic. Consider the following examples with the verb walk.
Those which are [+TELIC] allow inversion, whereas those which are
[-TELIC] do not.
(71) Walk you out of the door.
(72) Walk you into the garden.
(73) Walk you away.
(74) Walk you home.
(75) *Walkyou.
(76) *Walk you in the garden.
(77) *Walk you every day if you want to keep fit.
Telicity is the characteristic which determines unaccusativity in some
languages, for example Dutch (van Hout et al. 1993), and this suggests
that the class of verbs which takes imperative inversion in dialect A is
the unaccusative class.
54 BELFAST ENGLISH AND STANDARD ENGLISH
A further argument for this is the fact that precisely those verb
phrases which allow imperative inversion also allow inversion of the
following type, generally considered to be a characteristic of unac-
cusatives (on unaccusatives, see Burzio (1986), Levin & Rappaport
Hovav 1992, Rappaport Hovav & Levin 1992), and the discussion on
the analysis of dialect A imperatives below).
Many of the telic verbs are particle verbs in English—e.g., go away, sit
down, run off—so that very many inverted structures involve such
verbs.
Dialect B
In dialect B, inversion is possible with all verbs; it is not restricted to
the subset of verbs of motion which permit inversion in dialect A.
(96) Read you that book.
(97) Do you your best.
However, where the object is a full NP, it must occur after the
adverb.
(116) a. *Make you your mummy always a cup of tea.
b. Make you always your mummy a cup of tea.
Unlike other verbs, with which inversion is optional, where this verb
occurs with an overt subject it must precede the subject; inversion is
obligatory.
(124) a. Away you on.
b. *You away on.
(125) a. *Away youse ouside.
b. Youse away outside.
(126) a. Away you children to your beds.
b. *You children away to your beds.
Compare:
(130) She asked them to come see her.
(131) I told the children to go ask him.
(132) I go see my tutor every week.
Thus it is not simply the case that the verb away occurs in uninflected
contexts; rather, it is entirely restricted to occurring in imperatives.
Away cannot occur with a negative.
It would thus not be surprising to find that one of the varieties permit-
ted postverbal subjects. Pro-drop languages also appear to differ among
themselves in this way, with many pro-drop languages, for example
Italian (Rizzi 1982,1986), allowing postverbal subjects, whereas others
such as Brazilian Portuguese do not (Chao 1980). A rather neat analysis
could be provided by arguing that some characteristic of imperatives
(perhaps a strong AGR, necessarily marked as second person singular)
allowed them to be pro-drop, and that while Belfast English fell into
that group of pro-drop languages which permits postverbal subjects,
Standard English fell into the Portuguese-type group, which does not
permit these.
Unfortunately, however, the subject positions available in Belfast
English imperatives differ from those characteristic of postverbal NPs
in pro-drop languages. Thus in Belfast English imperatives, the VP-
final position characteristic of these preverbal NPs in pro-drop lan-
guages is not generally available; an inverted subject can occur VP-
finally if there are no other constituents in VP, but it cannot follow an
object or other constituent in the VP; although a vocative can occur
after the VP, as in standard English, a true subject cannot. Thus you can
only occur after the object of the verb where there is an intonation break
between the verb and you, indicating that it is a vocative rather than a
subject.
(150) a. Eat the apple, you.
b. *Eat the apple you.
c. Eat you the apple.
It may be objected that the presence of an obligatory intonation break is
difficult to perceive; however, there is a clearer difference between
postverbal NPs in the pro-drop languages and those in Belfast English
imperatives.
Overt-Subject Imperatives 65
This fact applies irrespective of whether the subject is the pronoun you
or a full NP.
The only case where the subject can occur after the object is where the
object is a weak pronoun, when the order verb-object-subject occurs.
Notice incidentally that in (164) and (165), the verb and object are not
adjacent, showing that the general adjacency requirement between verb
and object in English is not absolute and reinforcing the view that there
is no requirement for the verb and object to be adjacent at S-structure
for Case reasons.
It appears, then, that the verb has risen above adverbs, that is, out of
VP. However, in inverted imperatives, the verb may not precede nega-
tion.
(166) *Read you not that.
(167) *Be you not stupid.
This might seem to indicate that the verb raises to a node between nega-
tion and adverbs, as in French infinitives for example, where it has been
argued that the verb raises to AGR0, since it follows negatives and pre-
cedes adverbs.
(168) ne pas ecouter souvent la radio
not to listen to often the radio
Moreover, have, be, and auxiliaries cannot raise, even when an ele-
ment is required in C because of the presence of an inversion-triggering
element in SPEC/CP. This suggests that in standard English there is
some element in C already which blocks the raising of verbs into C, but
which, because it is not overt, is insufficient to satisfy the requirement
that an element appear in C.
(186) a. Nonfarlo.
not to do it
'Don't do it.'
b. Non lo fare.
it to do
'Don't do it.'
Kayne argues where clitics precede the infinitive, they are in fact
attached to an empty modal, rather than being left-adjoined to the infini-
tive; the presence of this empty modal in imperative infinitives (but not
other infinitives) explains the different order possibilities in imperatives.
Den Dikken argues that in Dutch the possibility of placing an object
NP on the right periphery of imperatives containing a resultative phrase
derives from the movement of an empty operator associated with the
object to the specifier position of a modal phrase. The right peripheral
position is available only in imperatives, and not in indicatives.
(187) Jan legde (die bal) neer (*die ball).
Jan put (that ball) down
(188) Leg (die bal) neer (die ball).
Put (that ball) down
that (183) and (184) are grammatical in this variety, as are examples
where the lexical verb has inverted.
(191) Under no cirumstances speak you to them.
(192) On no account open you that door.
suggesting that the verb in Belfast English has moved into C, thus satis-
fying the NEG criterion. The difference between Belfast English and
standard English then would be that in Belfast English, the element in C
is not an empty modal but an affix to which the verb can move.
There is further evidence of verb-raising in dialect B in the placement
of weak object pronouns. It is well known that in the Scandinavian lan-
guages, raising of the lexical verb triggers leftward movement of weak
object pronouns (see for example Holmberg 1986, 1992; Vikner 1991),
and it will be argued that this is also what is happening in Belfast Eng-
lish imperatives. The following example (from Vikner 1991) shows the
operation of object shift in Danish.
(193) a. I gar laeste Peter den uden tvivl ikke.
Yesterday read Peter it without doubt not
b. *I gar laeste Peter uden tvivl den ikke
c. *I gar laeste Peter uden tvivl ikke den
In order to show that object shift is happening in Belfast English
imperatives, we need to exclude the possibility that the object pronoun is
simply cliticising to the verb and raising with it. Two main factors show
that in fact this cannot be the case. First, although weak object pronouns
must precede adverbs in inverted structures, they may appear either
before or after the subject; the fact that they do not necessarily tag on to
the verb seems to indicate that they are not clitics. Note that the subject
itself need not be pronominal, so that attachment to the subject cannot be
analysed as attachment to an array of clitics attached to the verb.
(194) a. Read it everybody quickly,
b. Read everybody it quickly.
Second, there is evidence that weak pronouns are not clitics which
attach to verbs in the behaviour of such pronouns with the verb have
when it raises in English.
In British English (and Hiberno-English) have as a main verb may
raise to AGRS (and thence to C):
(195) a. I haven't any books.
b. I have always a lot of work.
c. Has John the answers?
72 BELFAST ENGLISH AND STANDARD ENGLISH
tion in English, a language which in general does not have main verb-
raising to C, where a dialect has a construction which involves the
movement of the verb to C. Speakers who allow inversion with all
verbs in imperatives, which we have analysed as verb movement to C,
all seem to have object shift; I have been unable to find any speakers
who allow widespread inversion but do not have object shift. In a sense,
it seems as if English has "latent" object shift, which in most varieties
does not show up because of lack of main verb-raising, but surfaces
when a dialect contains a construction involving main verb-raising.
We now turn to look at the position of the subject in dialect B imper-
atives. The question arises as to why the subject may either precede or
follow an object which has undergone object shift.
(211) a. Throw you us your end there.
b. Throw us you your end there.
(212) a. Find you it if you can.
b. Find it you if you can.
Conclusion
It has been shown here that there is more than one possible grammar in
relation to imperatives, depending on whether or not the verb raises,
78 BELFAST ENGLISH AND STANDARD ENGLISH
and whether or not the subject raises, before spell-out. These possibili-
ties are available within the Minimalist approach to parameter setting,
and are used in Belfast English but not in standard English.
Although the possibility of leaving the subject in situ can be seen to
relate to the general optionality of movement to SPEC/AGRSP in
Belfast English, the availability of verb-raising is restricted to impera-
tives, and thus is essentially construction specific; it involves a state-
ment about the strength of the V-feature of a particular morpheme in C,
the imperative morpheme, rather than about C in general. As noted in
Chapter 2, the grammar must be able to include statements about the
strength or weakness of particular items in a node, as well as general
statements about that node. Although this increases the power of the
grammar, it seems to be empirically necessary; moreover, given the
very strongly restricted range of options available within the Minimalist
program, it does not seem that it will make parameter setting unneces-
sarily difficult.
Consider the child, endowed with an innately specified Universal
Grammar, acquiring Belfast English. Assume that the child has set the
parameters affecting the position of the subject and verb in general, that
is, that the NP feature of Tense is strong (forcing subject-raising), but the
V-feature of Tense is weak, so that the verb generally remains within
VP. The child then encounters inverted imperatives, which do not fit into
the current grammar. There appear to be only a small number of possible
analyses available. Either the subject does not raise, meaning that the
NP-feature of some element is weak in imperatives; or the verb does
raise, meaning that the V-feature of some element is strong in impera-
tives. There is a very restricted choice of possible grammars, so that,
even if strength or weakness has to be specified for individual functional
elements rather than whole categories, the available possibilities are suf-
ficiently restricted for learning to take place easily.
One very interesting aspect of Belfast English imperatives is the
changes which appear to be in progress. The trend is approximately as
follows. For older speakers, widespread inversion is usually possible,
while among younger speakers dialect A prevails, and among many of
the youngest speakers—children under about 12— there is no inver-
sion, the grammar in relation to imperatives being like standard Eng-
lish. Now what is particularly noteworthy is that children learning
Belfast English, and whose parents presumably provide a significant
proportion of their input, often have grammars which do not admit
Overt-Subject Imperatives 79
Other such varieties include Ottawa Valley English (Carroll 1983) and
Ozark English (mentioned briefly in Chomsky & Lasnik 1977, Chomsky
1981, Roster & May 1982).
In standard English, for never appears immediately before to; for
appears before a lexical subject, but is absent where the subject is PRO.
The standard, pre-Minimalist, view of this (see for example Chomsky
1981) has been that the subject position of infinitives is ungoverned and
thus not Case-marked; it could therefore be PRO, considered to occur
only in ungoverned positions; if there is a lexical subject, this must be
Case-marked in another way, for example by for which governs and
Case-marks the subject position. Where for occurs, a PRO subject
should be impossible since the subject position is governed.
(3) a. For John to win would be amazing,
b. *John to win would be amazing.
(4) a. To win would be amazing.
b. *For to win would be amazing.
The status of PRO within the Minimalist framework is less clear, but
it seems likely that the restrictions on the occurrence of PRO can be
derived along the following lines: PRO is not Case-marked, and there-
81
82 BELFAST ENGLISH AND STANDARD ENGLISH
This is in fact the more common variety, and the majority of for to
speakers restrict their use to this type of clause. Finlay (1988), in a
study of the speech of Belfast schoolchildren, found examples of for to
only in purpose clauses, suggesting that the variety of Belfast English
featuring a wider use of for to may be restricted to older speakers and
hence is disappearing.
It is, however, to the less common, and more interesting, "strong" for
to variety, which uses for to as an alternative to to in the majority of
infinitive types, that we turn our attention here; this variety appears to
be restricted to Belfast and a few other areas, including south Armagh
and west Tyrone.1
In the strong for to variety of Belfast English, what is striking is that
for to can occur in infinitives in a range of different positions, contrary
to the data reported for other for to dialects, where usage seems to be
limited to those infinitives which are the complement of want-type
verbs and adjectives, which in standard English take for complements,
for example want and difficult. In addition to these environments,
Belfast English uses/or to infinitives in a variety of other cases.
Moreover, for to can be used even with verbs which do not take for in
standard English, and which are usually assumed not to be subcate-
gorised for the for complementiser.
However, the same construction occurs with the expletive there, which
is restricted to subject position, showing clearly that for to can indeed
occur after the infinitival subject
(25) I want there for to be some peace and quiet sometime.
(26) I'd hate there for to be ill-feeling.
Its failure to occur with for suggests that the for of for to is in fact the
complementiser itself. This is reinforced by the fact that it cannot co-
occur with whether, as shown in (15), repeated below.
(15) *I don't know whether for to go.
88 BELFAST ENGLISH AND STANDARD ENGLISH
Note that for to can in general occur with wh-words; thus sentences
like (34) and (35) are grammatical.
(34) I don't know where for to go.
(35) He wasn't sure what for to do.
This is exactly what we would predict if both whether and for are com-
plementisers; for can co-occur with wh-words which appear in the
(SPEC/CP) position, but not with another item which occurs in the
complementiser position itself. Thus, the lack of for to after whether
arises because the complementiser position is already filled by whether.
The suggestion that whether is a complementiser is, however, not
uncontroversial. In particular, it is in contradiction to the analysis pro-
posed by Kayne (1990) following suggestions by Larson (1985). Kayne
argues that whether occurs in SPEC/CP, and that this explains the con-
trast between whether (which allows the wh-infinitive construction) and
if, which does not, as shown in (36a & b) below.
(36) a. He doesn't know whether to go to the movies.
b. *He doesn't know if to go to the movies.
Although the evidence above suggests that the for of for to infinitives is
indeed the for complementiser, the questions raised in a-d, above, suggest
that it differs from the normal for complementiser in a number of ways.
lives with for to to occur, it fails to account for the other features of
Belfast English we have noted, namely, the restrictions on negative
placement and the occurrence of lexical subjects before for, for we
would expect to be able to get a negative like (49) if for is optionally a
governor, or optionally made Case-checking available.
(49) *For PRO not to go would be stupid.
whereas, as we have seen, this is ungrammatical; moreover, there is noth-
ing in the optional governing status of for that would lead us to expect to
find a lexical subject before it, as in (20) and (21) repeated below
(20) I wanted Jimmy for to come with me.
(21) I don't like the children for to be out late.
Note that it is not only the standard account of PRO that runs into dif-
ficulties here; thus, for example, Bouchard's (1983) proposal that it is
lack of Case rather than government which sanctions PRO also has
problems, since for is a Case assigner. Likewise, Borer's (1989)
account of infinitival subjects, under which these are not PRO but pro,
also encounters problems, for it requires an empty C slot into which
INFL can raise, and of course the complementiser position appears to
be filled in the for to construction.
Thus, we seem to have a contradiction. For appears to be a comple-
mentiser capable of governing/checking Case; but it also appears to be
able to occur with PRO. We can resolve this by claiming that, although
for is a complementiser, it may move out of the complementiser posi-
tion. More precisely, for can be a clitic in Belfast English, moving to
Tense and cliticising to to. This would not be the only example of com-
plementisers cliticising; thus Shlonsky (1988) argues that the Hebrew
complementiser se has clitic status, moving out of the complementiser
position and cliticising to INFL or the subject.
Let us explore the consequences of allowing for to cliticise to to. The
structure of a sentence like (9) above would then be:
(50) PRO for to stay here would be just as expensive.
I assume, essentially following Chomsky (1989,1992), that items
which do not enter into semantic interpretation at LF do not leave
traces, and that for is such an item. Thus, once for has moved, PRO is
ungoverned as required. The claim that for has no semantics is not
uncontroversial; a number of analyses of infinitives such as those by
Bresnan (1972), Stowell (1982) and Pesetsky (in preparation) have
92 BELFAST ENGLISH AND STANDARD ENGLISH
argued that for has a meaning, in particular, that for clauses are unrealised
with respect to the matrix verb. These accounts draw attention to the con-
trast between infinitives with for (or which can potentially take for) and
those which cannot take for. The former are necessarily "unrealised" with
respect to the matrix verb, whereas the latter are not, as the contrast
between (51) and (52) (with for) on the one hand and (53) and (54)
(which cannot take for in standard English) on the other serves to show.
(51) I want very much for John to win.
(52) I'd like very much for John to play the piano.
(53) I remember John to be the smartest.
(54) Bill considers himself to be intelligent.
However, all of these analyses are based on the observation that there
are certain types of infinitive with which for does not occur in standard
English—in particular raising and ECM verbs—and that the comple-
ments of those verbs are not—unlike other infinitival complements—
necessarily unrealised with respect to the matrix verb. But we have
noted that this distinction does not exist in Belfast English—all infini-
tives may have for, including those which are complements of raising
and ECM verbs, as (27a) and (28a) repeated below indicate.
(27) a. John seems for to be better.
(28) a. I believe them for to have done it.
Thus for does not seem to have a semantic contribution, since any
infinitive can occur with for. Therefore the observations about the
semantic content of for cannot hold. This means that for need not be
considered as having semantic content in this dialect, and therefore the
observations about the semantics of for in standard English noted above
do not contradict our claim that for does not leave a trace.5
Shlonsky (1988) also argues that a cliticised complementiser does not
leave a trace. For in Hebrew, constructions with the cliticised comple-
mentiser se do not show thai-trace effects:
(55) Ze ha- is se- ani na'amina, se- (hu) lo ohev salat xacalim.
this the man that I believe that (he) NEG like salad eggplants
'This is the man that I believe that (he) doesn't like baba ganouj.'
In (29b),/or has clearly cliticised to to, since it occurs after the infiniti-
val subject them; thus to must be the independent, non-clitic to, and so
cannot lower over the negative to affix to the verb. Similarly in (30b),
for has necessarily lowered under our analysis to leave PRO
ungoverned, and therefore to cannot be the affix and cannot move over
the negative to affix to V.
An alternative view of this, along Minimalist lines, would see to gen-
erated within VP and raising to Tense for checking. At spell-out, to
could be either still in the VP, below NEG (as in [a] or in T above NEG
(as in [b]). Since to has no semantic content, it, unlike verbs, should be
able to raise across NEG; since it will delete at LF, the improper chain
thus created will be invisible.
Now if in Belfast English, for can only cliticise to T, then we can see
why the sequence not for to is impossible; it would involve the cliticisa-
tion of for, not to T, but to an element in VP.
96 BELFAST ENGLISH AND STANDARD ENGLISH
Thus, the pre-for position of the subject cannot be the result of move-
ment; there is no reason why the subject should move into the postver-
bal object position, where nothing would be checked. The only move-
ment which should take place is movement to SPEC/AGR0.
As noted in the previous chapter, there is one instance where objects
move in the syntax to SPEC/AGR0P, and that is where the sentence is
imperative and the lexical verb raises to C. Here, an object which is a
weak pronoun moves to SPEC/AGR0. Let us look at what happens
when an infinitival complement is embedded under such a structure.
Verbs of the want class in general are not very felicitous in imperatives,
but contexts can be constructed in which they are possible.
(63) Want you for to win (and you will win).
(64) Expect you for to pass the exam (and you will pass).
Here, where there is a weak pronoun subject in the infinitive, it does
object shift.
(65) Want him you for to win (and he will win).
(66) Expect her you for to pass the exam (and she will pass).
For To Infinitives 97
Such adverbs can only occur before for and never between for and to
(68) a. I want the boys definitely for to be there,
b. *I want the boys for definitely to be there.
The structure of the (b) example would be the following, with a COMP-
trace sequence just as in (72).
(73) *WhOj do you think CP[ti that IP[ti will help you]
Note that CP-deletion must occur at some point in the derivation after
/br-cliticisation has applied, so that we find examples like (27a) and
(28a) where for has cliticised to to
(21 a) John seems for to be better.
(28a) I believe them for to have done it.
It would thus not seem to be the case that such verbs select IP; rather,
all verbs taking sentential complements select CP, and the deletion of
the CP node is an operation which takes place after certain verbs in the
course of the derivation.
Try-type Verbs
We have argued that several of the properties of Belfast English infini-
tives can be explained if for is a clitic. This accounts for a number of
facts about the dialect. However, it also raises some problems relating
to the subcategorisation of verbs taking infinitives.
One of these relates to the status of the difference between wanf-type
verbs, which can take infinitives with lexical subjects, and try-type
verbs, which cannot. This distinction holds in Belfast E glish, just as in
standard English
In analyses of standard English infinitives (see for example Chomsky
1981), it has often been claimed that the difference between these two
types is that the former take the for complementiser whereas the latter
do not, and thus do not have any means of Case-marking (or in more
recent frameworks, Case-checking) the lexical subject of an embedded
clause, which is therefore excluded. The occurrence of for with try-type
verbs in Belfast English is thus something of a problem. One would
predict that a dialect which allowed for with try would also allow lexi-
cal subjects with try, but as we have seen, that is not the case in Belfast
English. This indicates that the difference between want- and fry-type
verbs cannot lie in the selection of for, and that the difference must be
encoded in the lexicon by other means. One possibility is that that there
is a semantic difference between the two types. Pesetsky (in prepara-
tion), who claims for somewhat different reasons that fry-type verbs
take for, suggests that the difference lies in the fact that fry-type verbs
have agentive subjects whereas wanf-type verbs do not. A problem with
this analysis, however, is that fry can, as we have noted, be followed by
For To Infinitives 101
no link to a trace. Yet, clearly, lowering is not unbounded: for can only
lower to the nearest to.
Now, significantly, a number of proposals for lowering have involved
complementisers. As already noted, Shlonsky (1988) argues for comple-
mentiser cliticisation in Hebrew. McCloskey (1992a) argues for the low-
ering of the complementiser to I in Irish; as in Belfast English, comple-
mentisers in Irish show characteristics of being in C and of being in I.
Now complementisers have in common that they are often devoid of
semantic content. Let us suppose that lowering is possible to satisfy a
requirement to cliticise by PF, but will be impossible if a chain is
thereby created which will be ruled out at LF; only if the moved ele-
ment has no semantic contribution, and thus it and its trace are invisible
at LF, will lowering be possible.
As regards bounding, note that a "shortest movement" requirement
(Chomsky 1992), deriving ultimately from some kind of "least effort"
principle, is independent of direction of movement. It is not, as
McCloskey (1992a) notes, necessarily a condition on chains, but rather
a condition on the movement process itself. Thus, lowering will be con-
strained just as raising is, applying in the case of Heads on a head-to-
head basis.
Chomsky (1981) with respect to Ozark English but which could also
apply to the Ottawa Valley data, is that for is optionally a governor in
some varieties. Carroll notes that this can handle the Ottawa Valley
English data but regards it as unlikely that a relatively conservative
dialect should have acquired a property such as optionality of govern-
ment. Carroll's argument that for has been recategorised as something
other than a complementiser gains some weight from the fact that, in
the wea for to variety of Northern Irish English, which permits for to
only in rpose clauses, something similar seems to have happened; the
for here would seem not to be a complementiser but an item similar to
in order as in in order to.
It seems likely that, as the use of for to dies out, at some point for
becomes recategorised as something other than a complementiser,
which seems to have happened in rather different ways in Ottawa
Valley English and the weak variety of the for to dialect in Northern
Ireland.
This leaves us with Ozark English, where the data is as we have
noted somewhat unclear. Assuming that the paradigm is as reported in
Chomsky & Lasnik (1977), however, it seems that here for to, as in
Ozark, is restricted to cases where "for + lexical subject" is possible
in standard English (and also try, discussed above). The arguments for
Ottawa Valley English may thus apply there also.
In the light of the facts of Belfast English, other for to varieties need
to be re-examined; for example, we would wish to know how the use of
for to interacts with negative placement; however, from the data avail-
able, it seems likely that there is not a unitary for to phenomenon but
rather that this results from different factors in different dialects.
Conclusion
We have seen that Belfast English differs from standard English with
regard to the lexical status of for, in both varieties for is a complemen-
tiser, but only in Belfast English can it be a clitic.
In other for to varieties, the for to phenomenon seems to arise from
other lexical characterisations of for. In the "weak" for to dialect spoken
in Northern Ireland, for is an element similar to in order of in order to;
it is not a complementiser. In Ozark English, for is claimed to be a
104 BELFAST ENGLISH AND STANDARD ENGLISH
105
106 BELFAST ENGLISH AND STANDARD ENGLISH
That Belfast English has true embedded inversion can be seen from
the fact that a direct rendition of the words spoken or thought is not
required. Rather, embedded questions in Belfast English show
"sequence of tenses," as in indirect questions, with a verb spoken in the
present tense changed into the past when it appears embedded under a
past tense verb.
(9) a. She asked, "Are they leaving?"
b. She asked if they were leaving.
c. She asked were they leaving.
Thus, these are true indirect questions and not merely direct quotations.
Inversion in embedded questions is considered by most speakers to
be slightly better with yes-no questions than with wh-questions, and for
a not insubstantial number of speakers, inversion is only grammatical in
embedded yes-no questions. For these speakers, sentences such as
(12)-(15) are ungrammatical.
(12) She asked who had I seen.
(13) They wondered what had John done.
(14) They couldn't understand how had she had time to get her hair done.
(15) He didn't say why had they come.
This suggests that the process of inversion is not triggered in exactly the
same way in the two types of questions; it is possible to have a grammar
which permits inversion in yes-no embedded questions, but not in wh-
ones.
Inversion in Embedded Questions 107
As usual the standard English form does not allow inversion in the
embedded clause.
(40) Who did John hope (that) he would see?
(41) What did Mary claim (that) they stole?
It might be thought that sentences like (35) and (36) are not true
embedded clauses, but rather sentences containing a parenthetical ele-
ment, along the following lines.
Inversion in Embedded Questions 109
sion and complementisers. Thus inversion does not occur where a com-
plementiser is present.
(51) a. Did they read the book?
b. I wonder if they read the book.
c. *I wonder if did they read the book.
The word order in English questions is very similar to that found in
most main clauses in a number of Germanic languages, which are gen-
erally described as "Verb second" (V2) because, in main clauses, the
verb occurs in second position (see, for example, Vikner 1991); this
positioning of the verb is usually analysed as a requirement that the
verb move to C and another element move to the SPEC/CP position. In
general, this "verb second" order is restricted to main clauses; it cannot
occur in subordinate clauses (with some exceptions: for example Yid-
dish which, Diesing (1990) argues, allows embedded V2).
Although all main clauses in German and other V2 languages have
verb-raising to C, this only applies to certain sentence types in English:
questions, and certain sentences with sentence-initial negative-type
adverbs.
(52) Will they win the prize?
(53) Where will they go?
(54) Under no circumstances should they leave.
(55) Never have I heard such a thing.
Rizzi (1991) proposes that English has criterial verb-second; the verb
moves to C only when forced to by certain conditions, including the
Wh-criterion:
(56) a. Each +wh SPEC must have a +wh X°.
b. Each +wh X° must have a +wh SPEC.
(a corresponding NEG-criterion ensures verb-raising with negative
adverbial elements in SPEC/CP).
In English direct questions, the wh-element (if any) moves to
SPEC/CP and the verb moves to C, to satisfy the wh-criterion.
Although in English wh-elements front in embedded clauses, the verb
does not invert with the subject.
(57) a. I wonder where the children are going.
b. SE* I wonder where are the children going.
This might be thought to call the wh-criterion into question. However, it
has been suggested (see Rizzi & Roberts 1989) that there must be a
Inversion in Embedded Questions 111
This means that the lower CP is not directly selected by a lexical cate-
gory, and thus behaves like an unselected CP in relation to adjunction
Inversion in Embedded Questions 113
Compare (71):
(71) John claimed that under no circumstance would he do such a thing
arise from the same source as does the CP-recursion type inversion
found in the NEG-criterion examples.
Finally, note that a CP recursion analysis would predict that there
should be no difference in status between yes-no and wh-embedded ques-
tions in relation to inversion; if embedded CPs behave exactly like unse-
lected, matrix CPs, then inversion in both types should be equally possi-
ble. However, as observed above, many speakers find inversion better in
the yes-no type, and for some it is completely unavailable in the wh-type.
It thus seems that there are a number of arguments against considering
embedded inversion in Belfast English as deriving from CP-recursion.
Tense, rather than AGRS and in that case there was no agreement
between the subject and the verb. It was shown that where the subject
occurred in SPEC/Tense, it could not be a nominative pronoun; we
argued that this was so because in the SPEC/Tense position only default
Case could be checked.
(73) They is going.
(74) *We has read the book.
It is also worth noting that singular concord does not by any means
necessarily co-occur with embedded inversion in the grammars of
speakers. Almost all Hiberno-English speakers use embedded inver-
sion, and it is certainly in very widespread use in Belfast among speak-
ers of all classes, having the status of a local standard form; in fact most
speakers are not aware that it does not occur in standard English. How-
ever, it is only in certain areas (including Belfast) that singular concord
is used, and even there it tends to be absent from the usage of educated
middle-class speakers. There are thus very many speakers who do not
have grammars which admit the occurrence of an NP in the SPEC/TP
position in singular concord, but nevertheless have embedded inversion
in questions. Therefore, it does not seem likely that the availability of
embedded inversion is a consequence of the same parameter setting
(availability of TP as a subject-checking position) as singular concord.
An Alternative Analysis
Yes-No Questions
What we propose here is that there is in fact only one complementiser
position, in embedded questions in English, to which the verb moves in
Belfast English, but not standard English, before spell-out. In Belfast
Inversion in Embedded Questions 117
What the Belfast English facts suggest is that the trigger for verb
movement in embedded wh-questions is the presence of a [+wh] C, cre-
ated by agreement with a wh-element in SPEC/CP. This agreement
seems to fail in embedded questions in standard English, but to be oper-
ative in Belfast English. Further evidence for agreement in the overt
syntax in Belfast English comes from examples like (79)
(79) Who did you claim did he see?
In Belfast English, the verb can raise, provided C is not overtly filled,
but in standard English it may not. Now, the complementiser here must
initially be a [—wh] one, since claim does not subcategorise for [+wh]
complements.
(81) *They claimed if we won.
(82) *We claimed what we did.
And where an overt complementiser shows up, it is always that rather
than an interrogative complementiser.
But in Belfast English, the verb, which we have claimed to raise to +wh
complementisers, can raise here. How is this possible? It appears that
agreement takes place between the wh-element and the complemen-
tiser; and that agreement takes place in the overt syntax in Belfast Eng-
lish, but not until LF in standard English. There is thus a +wh-C in the
overt syntax in Belfast English, forcing the verb to move to C before
spell-out, but in standard English the complementiser remains [—wh]
until LF, and movement of the verb therefore does not happen until
after spell-out.
Notice that we cannot account for the inversion by requiring that all
CPs in a sentence which is a question contain inversion; the inversion
cannot go lower than the clause from which the wh-element is
extracted.
(84) a. *Who do you think did John convince did Mary go?
b. Who do you think did John convince that Mary went?
Inversion in Embedded Questions 119
Conclusion
Belfast English allows what have been termed by Doherty (1993) "sub-
ject contact relatives." These are what appear to be relative clauses
where the relativised element is a subject, and where the relative pro-
noun is not phonetically realised.
(1) I have a sister lives in Dublin.
(2) There are people don't read books.
(3) It's always me pays the gas bill.
124
Subject Contact Relatives 125
It is not simply the case that where a subject is relativised, Belfast Eng-
lish always allows a null subject. In fact in the majority of cases, an
overt relative pronoun or complementiser is required when the subject
is relativised, just as in standard English.
(10) *They were looking for the girl had won the prize.
(11) *The students have an exam next week are working very hard.
(12) *I lost the book gives an account of this.
Only head nouns in certain types of main clauses take null subject sub-
ordinate clauses. According to Doherty (1993), this class includes sen-
tences with existential there:
(13) There is/are some students never do any work.
(14) There's a short-cut takes you to the shops.
What these contexts appear to have in common is not the nature of the
head noun or the syntax of the clause in which it occurs. Rather, it is a
discourse factor they have in common: the matrix clause introduces a
new individual into the discourse, and the following clause states some-
thing about that individual.
In all those cases where subject contact relative clauses are possible,
there is an alternative with an overt pronoun.
(27) There's one woman in our street she went to Spain last year.
(28) I have one student he speaks four languages.
Note that these are not two sentences simply placed side by side; the
meaning is not the same as with separate sentences.
(29) There's one woman in our street. She went to Spain last year.
must mean that there is one, and only one, woman who lives in the
street, and that woman went to Spain last year. However, the reading
where the clauses form a single sentence is the same as that of the stan-
dard English; that is:
(30) There is one woman in our street who went to Spain last year.
Subject Contact Relatives 127
which does not imply anything about the total number of women who
live in the street, but only says that the total of whom it is true that they
both live in the street and went to Spain last year is one.
Such "resumptive pronouns" only ever occur in subject contact rela-
tives; as in standard English, they do not occur generally in relative
clauses.
(31) *Fm looking for the book (that/which) you recommended it.
(32) *John married the girl (thatAvho(m)) he met her on holiday.
(33) *I was talking to the lecturer (that/who) she takes the linguistics
course.
(34) *The students (who/that) they are taking French have an exam this
week.
Thus, apart from embedded null subjects, Belfast English does not
appear to have any other characteristics of pro-drop languages; there-
fore, the embedded null subject phenomenon does not arise because
Belfast English is a pro-drop language.
Embedded null subjects do not occur in all types of finite subordinate
clauses. For example, they do not occur in adverbial clauses
(44) *He ate his dinner after got home.
(45) *When arrived it was raining.
It might seem that these are that-clauses; indeed there are possible alter-
natives with that in the cases of (46) and (47).
(48) Mary was lucky that she got picked for the team.
(49) John was unlucky that he got caught.
But with "as well," a that clause cannot occur; only an infinitive is pos-
sible
(50) *Mary was as well that she took the other job.
(51) Mary was as well to take the other job.
It might seem that examples like (35a) and (36a) (repeated below) are
examples of raising structures where raising has taken place out of a
tensed clause, since there are somewhat equivalent sentences with an
expletive in subject position, as in (56) and (57).
(35) a. He was lucky didn't get caught.
(36) a. The children were as well took their chance when they got it.
(56) It was lucky that he didn't get caught.
(57) It was as well that the children took their chance when they got it.
However, null subject tensed clauses are not generally possible with
raising verbs.
(58) *John is likely will win.
(59) *The children seemed enjoyed the film.
Note that only the subject can be null; null objects are not permitted.
(60) *He was lucky Bill didn't catch.
(61) * John was unlucky Mary spotted.
(62) *Mary was as well Jill chose.
130 BELFAST ENGLISH AND STANDARD ENGLISH
(79) You know your sister, tell her to come and see me.
(80) See you, get that finished at once.
Thus, the null subject here does not seem to be a variable, but rather a
pronominal which may, like other pronominals, happen to be coindexed
with an element outside its clause. Therefore, unlike relative clauses,
which require a bound variable, topic structures do not seem to. The
null subject here seems to be of the type of the "root null subject" found
in languages which are not pro-drop (Rizzi 1991). Note that, like this
element, the null subject cannot occur in clauses lower than the highest
clause of the comment.
(89) a. See John, never helps his mother.
b. *See John, I know never helps his mother.
(90) a. You know your sister, sent me a postcard from her holiday.
b. *You know your sister, I hope sent me a postcard from her holiday
Thus, we claim that the so-called subject contact relatives are in fact
root-type clauses, able like roots to have null subjects. The null element
in subject position may (but need not necessarily) be coreferential with
an element in the topic, just as any pronominal may be coreferential
with another outside its clause.
Although in general the comment is about one noun phrase in the
topic, it also seems to be able to be a comment on other elements; for
example in cases such as the following, the comment appears to be
related to the adjective.
(91) John was lucky got away with it.
Conclusion
Thus, in Belfast English just as in standard English, where the subject is
relativised, an overt relative pronoun or complementiser is required.
What appear to be relative clauses in violation of this condition in fact
are structures of another type.
It remains to propose how, formally, Belfast English and standard
Subject Contact Relatives 135
English differ so that the former, but not the latter, allows topic-type
structures of the sort we have been discussing. These are not in fact
'topics' of the same type as found in standard English and discussed in
Lasnik and Saito (1992). In both standard English and Belfast English,
topicalisation to a position adjoined to SPEC/IP is possible.
(92) Newspapers, I really like.
(93) This pen, he writes his essays with.
Here, there is a gap in the site coreferential with the topic. On the con-
trary, gaps are impossible except in subject position in overtly intro-
duced topics.
(94) *See newspapers I really like.
(95) *You know this pen, he writes his essays with.
136
Conclusion 137
Chapter 2
1. Care is needed in obtaining intuitions on these cases. In particular, it is
important to obtain judgments from speakers who themselves use the -uns
forms, rather than from speakers who have regularly heard the forms but do not
use them. The latter appear to analyse the -uns forms as pronoun-noun combi-
nations like you guys and thus, while permitting singular concord, do not allow
the -uns forms in tag questions, or for the second occurrence of a referent.
Chapter 3
1. This means that raising does not take place to a position above NEG; in
this, Belfast English differs from the Scandinavian languages, where the object
moves to a position above negation. It may be that the difference is caused by a
different position for negation in the two language types—in NEG° in English,
but left-adjoined to VP like adverbs in the Scandinavian languages.
2. Beukema and Coopmans (1989) argue that the null subject in imperatives
is a variable associated with a discourse-identified null topic, rather than pro.
However, there is evidence against this analysis in that a null subject is an
imperative cannot be third person when indentified by an appropriate discourse
element.
(i) Everybody take out their books. After that, write down their names
The second sentence cannot be interpreted to mean that everyone should write
down their own name; rather everyone must write down the names of another
group of people. The sentence must be interpreted as having the subject you,
rather than everyone. If the null subject were a variable bound by a discourse-
identified topic, then it ought to be able to be interpreted as everyone. On the
139
140 Notes
contrary, the null subject must be interpreted as second person, suggesting that
imperatives have an obligatorily L+2sg] AGR which identifies the subject.
Chapter 4
1. It appears that this dialect was more widespread in the relatively recent
past; thus Joyce (1910), in his book English as we speak it in Ireland notes that
" 'For' is constantly used before the infinitive," quoting as an example the fol-
lowing lines from a folksong:
exactly like the standard English for, and one which has no semantic content
and may appear elsewhere; there would, however, appear to be no arguments in
favour of this from the Belfast English data, where for appears to behave in the
same way in all cases.
Chapter 5
1. Wonder has another use in Belfast English; it can mean "to be surprised,"
and with this meaning takes a that complement.
(ix) *It depends which theory that makes the best predictions,
(x) *I wonder which author that wrote this book.
On the contrary, in the emphatic structure with it is/was, that occurs.
Baker, CL (1991) The Syntax of English not: the Limits of Core Grammar. Lin-
guistic Inquiry 22 (387-429).
Bayer, J (1984) COMP in Bavarian Syntax. The Linguistic Review 3 (209-274).
Beukema, F & Coopmans, P (1989) A Government-Binding perspective on the
imperative in English. Journal of Linguistics 25 (417-436).
Bobaljik, J & Jonas, (1992) Subject positions and the role of TP. Paper pre-
sented at the Glow Colloquium, Lund, Sweden.
Borer, H (1989) Anaphoric AGR. In Jaeggli, O & Safir, K (eds) (1989).
Bouchard, D (1983) On the content of empty categories Foris, Dordrecht, the
Netherlands.
Bresnan, J (1972) The theory of complementation in English syntax. PhD dis-
sertation, Massachusetts Institute of Technology, Cambridge, MA.
Bromberger, S & Halle, M (1989) Why Phonology is different. Linguistic
Inquiry 2Q (51-10).
Burzio, L (1986) Italian Syntax: a Government-Binding Approach. Reidel, Dor-
drecht, the Netherlands.
Carroll, S (1983) Remarks on FOR-TO infinitives. Linguistic Analysis 12
(415-451).
Chao, W (1980) PRO-drop languages and non-obligatory control. University of
Massachusetts Occasional Papers in Linguistics 1 (46-74).
Chomsky, N (1981) Lectures on Government and Binding, Foris, Dordrecht, the
Netherlands.
(1986) Knowledge of Language. Praeger, New York.
(1989) Some notes on economy of derivation and representa-
tion. MIT Working Papers in Linguistics 10.
(1992) A Minimalist Program for Linguistic Theory MIT Occa-
sional Papers in Linguistics 1.
Chomsky, N & Lasnik, H (1977) Filters and control. Linguistic Inquiry 8
(425-504).
142
References 143
Acquistion and dialect variation, 78-80, Embedded verb second. See Verb second,
137-138 embedded
Adverbs, position of, 19, 25-26, 57, 60, Exceptional Case Marking. See ECM verbs
67,71,97 Existential there, 122
AGRS, 21-23, 25, 29, 34, 42, 44, 62, 115;
see also SPEC/AGRS Faroese, 38
Agreement, 16^4 Finlay, 8, 16
Arabic, 21-22 for-to infinitives, 81-104
away, as a verb, 58-59 French, 45-46,61
Full copy, 120
Beukema, 48
Bobaljik, 30 German, 61, 68
Grammaticality judgements. See Intuitions
C, see Complementiser Gueron, 36
Carroll, 102-3
Case, 23-25, 28, 32-42, 44, 84, 91, 93 Head Movement Constraint, 42
Chinese, 133-34 Hebrew, 88
Clefts, 125 Highest subject restriction, 131
COMP-trace effect, 97 Historic present, 18
CP recursion, 111-15 Holmberg, 38, 71
Complementiser, 42-44, 66-76, 78,
88-89,97, 107, 109-15, 117- Icelandic, 111
123 Imperatives, 45-80, 96
lowering, 101-2 Infinitives, 62, 69, 81-104
Co-ordination, 23-24, 71 Intuitions, 12-13, 14, 24, 56
Coopmans, 48 Inversion,
in imperatives, 45-80
Danish, 113 in questions, 42-44, 88-89, 105-23,
Demonstratives, 18, 33-34 133
Den Dikken, 69-70 Irish, 7, 37, 119
Diesing, 115 Italian, 69-70
Discrimination against dialect users, 7
rfo-support, 32, 43, 49, 62, 66, 68 Jonas, 30
Doherty, 13, 124, 130-34
Duffield, 13, 28-29 Kayne, 32, 35,69
Dutch, 61,69-70 Koopman, 37
147
148 Index