Transcript NyePresentationFalkAuxiliaries

Functional relations in the
English auxiliary system
by Yehuda N. Falk
presented by Rachel Nye
Contents
1.
2.
3.
4.
5.
6.
7.
Auxiliaries: the issue
Supportive DO
Perfective HAVE
Progressive BE and passive BE
The modals
Falk’s conclusions
Final remarks
1. Auxiliaries: the issue
“There are two main ways to analyze auxiliaries. On the
one hand, an auxiliary can be treated as a special type
of raising verb, which has two arguments, a SUBJ and
an XCOMP…On the other hand, auxiliaries can simply
be considered as feature carrying elements, which
contribute tense/aspect or voice information to the
clause, but which do not have a PRED value or
subcategorization frame.”
(A Grammar Writer’s Cookbook (1999: 60-61) Butt, King,
Niño and Segond)
-Falk illustrates these two possibilities for the functional relation
between auxiliaries and the verbal elements which follow them with
this example:(1) a. The children will take syntax.
b. will is the head of the sentence, and [take syntax] is a
complement of will [aux-predicate]
c. take is the head of the sentence, and will is a ‘‘modifier’’ or
morphological marker expressing/realizing future tense [aux-feature]
(2) a. aux-predicate
(2) b. aux-feature
SUBJ [1]" the children "

PRED ' will  XCOMP  SUBJ'




SUBJ [1]


PRED ' take  SUBJ, OBJ ' 
XCOMP




OBJ " syntax "
 




SUBJ [" the children " ]

TENSE FUT



PRED ' take  SUBJ, OBJ '


OBJ
["
syntax
"
]


-Falk claims that most LFG accounts choose either the
aux-predicate or the aux-feature analysis for all
auxiliaries.
- Whilst this is perhaps something of an oversimplification, it is certainly true that even for those
accounts which do distinguish aux-predicate from auxfeature, the division is no more fine-grained than modal
vs. non-modal auxiliaries (see Table 1).
- Falk, however, argues that each auxiliary must be
considered individually to determine whether it is of the
type aux-predicate or aux-feature.
Table 1: An overview of LFG approaches to auxiliaries
Account
Aux-predicate
Aux-feature
Bresnan2001)
(
None
All
Butt, King, Niño &
Segond1999)
(
All modals
All non-modal
auxiliaries
Dalrymple2001)
(
Possibly the
modals?
All non-modal
auxiliaries
Falk1984)
(
All auxiliaries
except supportive
DO
Supportive DO
Falk2008)
(
Progressive BE
All the modals
except WILLand
(
SHALL)and
WOULD
Supportive DO
Perfective HAVE
Passive BE?
WILLand
( SHALL)
and WOULD
2. Supportive DO (aux-feature)
(3) a. The children took syntax
b. The children did take syntax.
SUBJ [" the children " ]


(4) TENSE PAST


PRED ' take  SUBJ, OBJ '


OBJ
["
syntax
"
]


(5) a. did:
b. take:
c. took:
TENSE = PAST
PRED = ‘take <SUBJ, OBJ>
PRED = ‘take <SUBJ, OBJ>
TENSE = PAST
-The use of the synthetic form (3) a. over the analytic form
(3) b. is, according to Falk, a simple reflex of ‘the Economy
of Expression Principle, which…prefers morphological
expression to syntactic expression’.
-As such, ‘There is no reason to think that DO itself has a
focusing function, or that the emphasis facts are a
“theoretically central issue” which must be expressed as a
lexical property of DO.’
-Falk (1984) concludes that DO is a feature-carrier, not a predicate
‘on the basis of co-occurrence restrictions’
use to
(6) TENSE =c PAST
[i.e., the clause in which used to appears must have the past tense
feature]
(7) a. The children used to take syntax.
b. The children did not/so use to take syntax.
c. Did the children use to take syntax?
- did in (7) b. and (7) c. plays the same role that the past tense
suffix –d plays in (7) a - i.e. a morphological feature-carrier.
-If it were a separate PRED, then the condition in (6) would not be
met.
3. Perfective HAVE (aux-feature)
-
‘[perfective] HAVE provides two features, a tense feature and an
aspectual feature. The past participle form, on the other hand,
provides the verb’s predicate but no tense or aspectual information.’
(8) have
TENSE = PAST
ASP = PERF
(9) a. The children have taken syntax.

(9) b. SUBJ [" the children " ]
TENSE PRES



ASP PERF



PRED
'
take

SUBJ
,
OBJ

'


OBJ [" syntax " ]

-
Evidence that have is the (aspect) feature carrier:-
(i) The participle is not sufficient for the perfective aspect
(10) a. Anyone taking syntax [=anyone who is taking syntax] should
have his head examined. [present participle]
b. Any subject taken by the children [=any subject which is
taken by the children] is boring. [passive participle]
c. *Anyone taken syntax [=anyone who has taken syntax]
should have his head examined. [past participle]
(ii) And it isn’t even always necessary
(11) A: (Taking orders for lunch.) Who eats falafel?
B: Well, I have in the past, but I really don’t want any now.
-
There are also co-occurrence restrictions (parallel to the situation
with DO and used to)
better
(12) PRED = ‘better <XCOMP> SUBJ’
TENSE =c PAST
ASP =c PERF
(13) The children had better take syntax.
SUBJ [1]" the children "

TENSE PAST



ASP PERF



PRED ' better  XCOMP  SUBJ'


SUBJ [1]





 XCOMP PRED ' take  SUBJ, OBJ ' 

OBJ " syntax "
 




- If HAVE had a predicate, the constraints in (12) would not be met.
(14) If the verb has a c-structure complement, it belongs
to the category VP[part].
The constraint in (14), which Falk suggests forms part of
the lexical entry of forms of HAVE, ensures that when
perfective HAVE selects a complement, this complement
takes the form of a past participle.
4. Progressive BE (aux-predicate) and
passive BE (aux-feature?)
-Progressive BE is often grouped together with perfective HAVE, but Falk
suggests that there are reasons to believe this is misguided:(15) a. The children were taking syntax.
b. The children started taking syntax.
c. The children kept taking syntax.
d. The children stopped taking syntax.
(16) a. The children were in the state of taking syntax.
b. The children entered the state of taking syntax.
c. The children continued in the state of taking syntax.
d. The children left the state of taking syntax.
‘It is the participial complement which is progressive in these examples, the
governing verbs specify the relationship between the subject and the
progressive state’.
Thus it is the participle and not progressive BE which is specified as:(17) ASP = PROG
The f-structures for (15) a. The children were taking syntax and (15) b. The
children started taking syntax are thus very similar, as represented below in
(18)a. and (18)b. respectively:(18)a.
SUBJ [1]" the children "

TENSE PAST



PRED ' be  XCOMP  SUBJ'



SUBJ
[
1
]




ASP PROG



 XCOMP 
PRED ' take  SUBJ, OBJ ' 






OBJ
"
syntax
"









(18)b.
SUBJ [1]" the children "

TENSE PAST



PRED ' start  XCOMP  SUBJ'



SUBJ
[
1
]




ASP PROG



 XCOMP 
PRED ' take  SUBJ, OBJ ' 






OBJ
"
syntax
"









Progressive BE is no different to main verb BE, which also expresses a state.
It can even be co-ordinated with other uses of BE.
Evidence in favour of a bi-clausal analysis of progressive BE:(i) The participle complement of BE as well as verbs such as START, KEEP
can be replaced by a PP predicative complement:
(19) a. John kept Bill running/at a run.
b. Moe went on working/with his work.
c. Rodgers is working/at work on a new play.
(ii) It is possible to have distinct modifiers for each of the clauses:
(20) Today, the workman is coming tomorrow (but tomorrow that may
change).
The f-structures in (21) a. and (21) b., showing an aux-predicate and an auxfeature analysis respectively make clear why the former is necessary.

(21) a. SUBJ [1]" the repairman "
(21) b. SUBJ [" the repairman " ]
TENSE PRES



PRED ' be  XCOMP  SUBJ'



ADJ " today "


SUBJ [1]


ASP PROG




 XCOMP PRED ' come  SUBJ ' 







ADJ
"
tomorrow
"







TENSE PRES



ASP PROG



PRED
'
come

SUBJ

'




 " today " 
ADJ 


["
tomorrow
"








This analysis of progressive BE also accounts for the word order restrictions
shown in (22) a. and b. :(22) a. The children have been taking syntax.
b. *The children are having taken syntax.
The f-structure for (22) a. is well-formed, as shown in (23) a. In contrast, the
f-structure in (23) b. is ill-formed because the present participle having
specifies two conflicting values for the feature ASP – PERF as it is a form of
HAVE, and PROG because it is a present participle form, as (23) b. shows.
(23) a. well-formed
SUBJ [1]" the children "

TENSE PRES



ASP PERF



PRED ' be  XCOMP  SUBJ'


SUBJ [1]


ASP PROG




 XCOMP PRED ' take  SUBJ, OBJ ' 





OBJ
"
syntax
"







(23) b. ill-formed
SUBJ [1]" the children "

TENSE PRES



PRED ' be  XCOMP  SUBJ'



SUBJ [1]


ASP PROG







XCOMP
ASP
PERF




PRED
'
take

SUBJ
,
OBJ

'



OBJ " syntax "
 







Passive BE
-Unclear whether it should be analysed as aux-predicate or
aux-feature.
-Passive BE should be treated as a separate predicate just
as passive GET (presumably) is?
- But ‘this putative predicate seems impossible to isolate’ –
cannot be modified separately, doesn’t have a ‘be in state’
reading and can’t be co-ordinated with other complements
of BE.
- Hence passive BE may just be a way to provide tense.
- Falk states that more research is needed before the status
of passive BE can be confirmed, but argues that the current
state of the evidence favours an aux-feature analysis.
5. The modals
WILL and WOULD (aux-feature)
(24) a. The children did take syntax. (past tense)
b. The children do take syntax. (present tense)
c. The children will take syntax. (future tense)
d. The children would take syntax. (conditional tense)

- WILL and WOULD are simply tense carriers.
- Falk claims that there is evidence for the aux-feature status of
WOULD from co-occurrence restrictions similar to those discussed
for DO and HAVE:(25) a. The children would rather take syntax.
b. I would rather the children take syntax.
c. rather PRED = ‘rather <SUBJ, COMP>
TENSE =c CONDIT
Other
modals (aux-predicate)
(26) The children may take syntax.
a. ≈ It is possible that the children will take syntax.
(epistemic) POSSIBILITY
b. ≈ The children are permitted to take syntax.
(root) PERMISSION
-The fact that the modal can be paraphrased with the
predicates possible and permitted, suggests that an auxpredicate analysis for the modal itself is plausible.
- Stronger evidence for a predicate analysis comes from the
fact that in the root usage, the modal places selectional
restrictions on its subject which is a thematic argument of
the modal. (27) below, with a pleonastic subject, can only
have an epistemic reading:(27) There may be children taking syntax. (≈ It is possible
that there are children taking syntax.)
- Modals such as DARE, which have only a root use, cannot
occur with pleonastic subjects:-
(28) a. The children dare not take syntax.
b. *There dare not be any children taking syntax.
- For the modals analysed as aux-predicate (i.e. all those besides
WILL and WOULD), root modals are taken to be equi (control)
predicates, whilst epistemic modals are raising predicates.
-The f-structures corresponding to the two readings of (26) The
children may take syntax are given below in (29) a. and b:-
(29) a.
(29) b.
SUBJ [1]" the children "

TENSE PAST



PRED ' mayroot  SUBJ, XCOMP '



SUBJ
[
1
]






 XCOMP PRED ' take  SUBJ, OBJ ' 



OBJ " syntax "
 

SUBJ [1]" the children "

TENSE PAST



PRED ' mayepist  XCOMP  SUBJ'



SUBJ
[
1
]




 XCOMP PRED ' take  SUBJ, OBJ ' 



OBJ " syntax "
 

- Falk admits, ‘For the epistemic modals, the argument for an auxpredicate analysis is weaker, since there is no relation of selection
between the modal and the subject’.
- Nevertheless, there is some evidence which inclines one to such a
view:(i) Separate modification of the two clauses is acceptable for some
speakers
(30) ??Today, the repairman may come tomorrow (but tomorrow that
may change).
(ii) Adverbials can modify the verb or the modal, as the differences in
grammaticality/interpretation below show
(31) a. *Tabs never may be kept on syntax students.
b. Tabs may never be kept on syntax students.
(32) a. Tabs never should be kept on syntax students.
(≈There is never an obligation.)
b. Tabs should never be kept on syntax students.
(≈There is an obligation to never keep tabs on them.)
Epistemic modals
- There is at least some weak evidence in favour of an aux-predicate
analysis even for epistemic modals.
- No positive arguments in favour of the alternative aux-feature analysis
e.g. no co-occurrence restrictions.
Root modals
- Must be aux-predicate.
- There is no alternative analysis possible for root modals, given that
they thematically select their subjects.
- So even if an aux-feature analysis is later found to be preferable for
epistemic modals, this will not influence the status of root modals.
6. Falk’s conclusions
-
-
-
There is no single analysis that covers all auxiliaries.
Some must be aux-feature whilst others are auxpredicate.
Most other accounts (wrongly) assume one analysis
across the board for all auxiliaries.
LFG, with the distinct levels of c-structure and fstructure, is particularly well-suited to capturing the
functional diversity of auxiliaries, despite their similarities
in constituency.
7. Final remarks


Falk includes quite a lot of data from other languages,
which I haven’t discussed here. Despite explicitly stating
that analysis of an auxiliary as aux-predicate or auxfeature in another language does not entail anything
about the correct analysis for English, I find it hard to
believe he would include it if he didn’t think it
strengthened his case.
His dismissal of (potential) differences in the use of
auxiliaries in dialects other than his own (at least some
of which are dismissed as “substandard varieties”)
seems unhelpful, and somewhat surprising, given that
his goal is to give a finer-grained analysis of the
auxiliaries than other accounts have offered.