UCLA Working Papers in Linguistics, no.

12, September 2005

Proceedings of AFLA XII, Heinz & Ntelitheos (eds.)

SYNTAX AND SEMANTICS OF TAGALOG MODALS

A NNA H OLT
MIT
[email protected]

A LYA A SARINA
MIT
[email protected]

In this paper, we analyze modal verb constructions in Tagalog, which provide support for the theory
of modal proposed by Brennan (1993). There is a clear relationship between the syntax and semantics
of Tagalog modal verbs. In particular, we show that Tagalog modals assign a -role to the subject if,
and only if, they syntactically mark the subject. It follows that directed deontic modals, which assign
a -role to the subject, are syntactically different from non-directed deontic modals, which do not. We
show that dynamic modals preferentially assign a syntactic marking and a -role to the subject. An
analysis of all Tagalog modal verbs is presented.

1. T HEORIES

OF

M ODALS

There is an open debate in the literature as to how semantic meanings of modal verbs should
be classified and whether these semantic distinctions map onto underlying syntactic differences.
English is a difficult language in which to further this topic, as the potential syntactic differences
are not reflected in the phonology. However, overt syntactic markings in Tagalog allow us to argue
for one of the three major proposals for how syntax and semantics interact in sentences with modal
verbs.
Tagalog evidence indicates that the following modal classification scheme (c.f. Hall (2001))
reflects linguistically significant differences between modals:
(1)

Modals

Epistemic

Root
Deontic
Directed

Dynamic

Non-Directed

Thanks to Genara Banzon for spending many hours answering our questions about Tagalog, to Oly Fernando for
helping resolve some last-minute questions about the language, to David Pesetsky for help with our presentation and
our paper, and to Norvin Richards for everything he taught us. Thanks also to Kai von Fintel, and the audiences at
MIT and AFLA XII for their comments.

UCLA Working Papers in Linguistics, no. 12

These categories are defined in terms of the semantics of the modals. Epistemic modality refers
to speaker knowledge, e.g. It might rain tomorrow. Root modals refer to the state of the world.
They can be further subdivided into deontic verbs, reflecting permission or obligation, and dynamic verbs, which reflect ability (e.g. I can run a seven-minute mile.). Furthermore, there is a
key distinction between directed1 and non-directed2 deontic modals. Directed modals assign permission or obligation to the subject, as in You must do your homework. Non-directed deontic
modals do not attribute the permission or obligation to the subject in particular, as in The bread
must be eaten.
Traditional accounts of modality (e.g. Jackendoff (1972)) focus on the differences between root
and epistemic modals. It is argued that root modals correspond to a control structure and assign a
-role to the subject. Epistemic modals correspond to a raising structure and do not assign a -role
to the subject. For example:
(2)

a.

Youi must [PROi do your homework.]

b.

Iti might [ti rain tomorrow.]

(directed deontic)

(epistemic)

More recently, it has been argued by Bobaljik and Wurmbrand (1999), among others, that there
are no syntactic differences between root and epistemic modals. The fact that non-directed deontic
modals do not assign a role to the subject, and therefore appear to have a raising structure, has
been used to support this claim.
(3)

a.

The breadi must [ti be eaten.]

b.

Iti might [ti rain tomorrow.]

(non-directed deontic)
(epistemic)

Brennan (1993) argues for a middle ground. She proposes that non-directed deontic modals are
syntactically the same as epistemic modalsthey take a sentential complement, do not assign a
-role to the subject, and result in a raising structure. Directed deontic and dynamic modals, on
the other hand, take the subject and VP as arguments, assign a -role to the subject, and result in a
control structure.
Facts from Tagalog generally support this account. Tagalog makes clear distinctions between
directed and non-directed deontic modals, along the lines argued for by Brennan (1993). The
modal that is preferred for the dynamic reading assigns case (and a -role) to the subject. (The
dynamic reading is also allowed with modals that do not assign case to the subject.) Tagalog has
1
2

Sometimes called ought-to-do.

Sometimes called ought-to-be.

Asarina & Holt, Syntax and Semantics of Tagalog Modals

no epistemic modal verbs, but since all accounts agree that epistemic modals take raising structures,
there is no need for further evidence to distinguish between the theories.

2. ANG/NG M ARKING

IN

TAGALOG

Verbs in Tagalog bear morphology indicating which is mirrored by morphology on the NPs in the
sentence. Subject-ANG verbs (glossed as S.A. throughout the paper) cause common-noun subjects
to be introduced with the morpheme ang, glossed throughout as ANG. Object-ANG verbs (glossed
as O.A.) assign an ANG-marking to the direct object. NPs not marked with ANG are marked with
SA (dative) or NG (unmarked). In simple sentences containing a subject and a direct object,
one of the two NPs will generally be ANG-marked and the other NG-marked. For example:
(4)

a.

Bumili
ang lalaki ng kotse.
S.A-bought ANG man NG car
The man bought a car.

b.

Binili
ng lalaki ang kotse.
O.A.-bought NG man ANG car
The man bought the car

Note that modal verbs are object-ANG verbs, though they lack the morphology that appears on
ordinary object-ANG verbs.
(5)

a.

Kailangan ng lalaki ang kotse.

need
NG man ANG car
The man needs the car.

b. *Kailangan ang lalaki ng kotse.

need
ANG man NG car

3. K EY FACTS

ABOUT

TAGALOG M ODAL S ENTENCES

In Tagalog, the subject normally receives an ANG or NG marking from the lower verb. However,
in sentences with modal verbs, the subject can sometimes be NG-marked even though the lower
verb is a subject-ANG verb. Since the NG marking on the subject is not assigned by the lower
verb, it appears that the marking is assigned by the modal. This is consistent with the observation
that modal verbs mark their subjects with NG, as in (5).

UCLA Working Papers in Linguistics, no. 12

We will now consider four possible structures for modal sentences with S.A. lower verbs. All
Tagalog modal verbs allow the subject to appear before or after the lower verb. Some modals
require that the subject be ANG-marked (regardless of position), some require that the subject be
NG-marked (also regardless of position), and some, for instance kailangan (must), allow both
possibilities.
(6)

(7)

NG-marked subject:
a.

Kailangan ng lalaki bumili ng kotse.

must
NG man S.A.-buy NG car
The man must buy a car.

b.

Kailangan bumili ng lalaki ng kotse.

must S.A.-buy NG man NG car

(control)

(transparent clause)

ANG-marked subject:
a.

Kailangan bumili ang lalaki ng kotse.

must
S.A.-buy ANG man NG car
The man must buy a car.

b.

Kailangan ang lalaki bumili ng kotse.

must ANG man S.A.-buy NG car

(opaque clause)

(raising)

There is a semantic difference between the sentences in which the subject is NG-marked by the
modal and the sentences in which it is not. This difference provides further evidence that the
subject is indeed being NG-marked by the modal verb. In modal sentences with a subject-ANG
lower verb, the subject receives a -role from the modal if, and only if, it is NG-marked (by the
modal).3 As we will see, NPs that receive a -role from the modal must be interpreted as bearing the
obligation or ability described by the modal. A subject that pragmatically cannot receive obligation
from the modal therefore cannot be NG-marked by the modal:
(8)

a. #Kailangan ng tinapay mawala.

(control)
must
NG bread disappear
#The bread must make itself disappear.
b. #Kailangan mawala ng tinapay. (transparent clause)
#The bread must make itself disappear.

3
With O.A. lower verbs in modal sentences, it is of course ambiguous whether the subject is NG-marked by the
lower verb or by the modal. The subject can never be ANG-marked in these cases.

Asarina & Holt, Syntax and Semantics of Tagalog Modals

(9)

a.

Kailangan mawala ang tinapay.

must
disappear ANG bread
The bread must disappear.

b.

Kailangan ang tinapay mawala.

The bread must disappear.

(opaque clause)

(raising)

We now discuss the syntax of each of the four possible constructions in greater detail.
4. S YNTACTIC S TRUCTURES

FOR

TAGALOG M ODAL V ERBS

4.1. Control
The first type of sentence to consider is that in (6a), where the subject appears before the lower
verb and is NG-marked despite the lower verb being S.A.:
(10)

Kailangan ng lalakii [ CP (na) bumili PROi ng kotse.]

must
NG mani [ CP (LI) S.A.-buy PROi NG car]
The man must buy a car.

(10) optionally contains na, which is one of the forms of the Tagalog linker (glossed as LI); the
linker also appears as -ng. The relevant function of the linker is to indicate the left edge of a
clause.4 The surface position of the subject in (10) is thus in the higher clause, outside of the lower
CP. Since the subject is marked by and receives a -role from the modal, the subject is generated
in the higher clause and controls a PRO in the lower clause, as shown above. Kailangan thus takes
two arguments in the control structure: a subject NP, and a CP clause.
4.2. Transparent Clause
It is also possible for the subject in a modal sentence to appear after the S.A. lower verb, but be
NG-marked, as in (6b) (repeated here).
(11)

Kailangan [ VP bumili ng lalaki ng kotse.]

must
[ VP S.A.-buy NG man NG car]
The man must buy a car.

4
The linker can sometimes be left out in colloquial speech, and our main informant generally allowed for the linker
to be dropped.

UCLA Working Papers in Linguistics, no. 12

As in the control structure, the subject in the transparent clause structure receives a role and
a NG-marking from the modal. Since the subject in (11) appears in the lower clause, we must
explain why the modal is not blocked from NG-marking the subject. One possibility to consider is
that the lower verb raises to the modal, as suggested in Mercado (2002) (though Mercado (2005)
argues against verb raising). However, a particle can be inserted between the modal and the lower
verb in (11), which shows that verb raising does not occur in (11):
(12)

Kailangan pala
bumili ng lalaki ng kotse.
must
SURPRISE S.A.-buy NG man NG car
Oh, I see the man must buy a car.

A different possibility, one that appears to be correct, is that restructuring is taking place. That is,
the lower clause is not a full CP, but is reduced to a smaller clause, such as a VP. As part of her
argument that restructured clauses are functionally reduced, Wurmbrand (2001) notes that such
clauses cannot contain negation. And indeed, negation can be inserted between the modal and the
lower verb when the subject is ANG-marked (and the lower clause is not reduced), but not when
the subject is NG-marked and the lower clause is reduced:
(13)

a. *Kailangan(ng) [hindi bumili ng lalaki ng kotse.]

must-(LI)
not S.A.-buy NG man NG car
b. Kailangan(ng) [hindi bumili ang lalaki ng kotse.]
must(-LI)
not S.A.-buy ANG man NG car
The man must not buy a car.

Following Wurmbrands proposal, we expect the sentence containing the restructured clause to
look like this:
vP

(14)
DP
ng lalaki
NG man

v
v

VP
VP

V
kailangan
must

DP

bumili
buy

ng kotse
NG car

Asarina & Holt, Syntax and Semantics of Tagalog Modals

As desired, there is no attachment site for negation between the modal and the lower verb. Unfortunately, this structure does not correspond to the surface Tagalog word order. We thus propose
the following modified structure for the transparent clause modal sentences in Tagalog:

(15)

XP

Kailangan
X
Must

vP
DP
ng lalaki
NG man

v
VP

v
V

DP

bumili
buy

ng kotse.
the car.

This structure is similar to the one corresponding to Wurmbrands proposal. The clause that the
modal takes is not a full CP, though it is slightly larger than a vP. The extra attachment site allows
the lower verb to move up to precede the subject, without creating an attachment site for negation.
The subject stays in situ in the specifier of vP. This proposal accounts for the Tagalog word order,
the lack of attachment site for negation between the modal and the lower verb, and the fact that the
lower clause is transparent, i.e. allows the modal to assign marking to the subject. However, while
the proposed structure appears to be the most natural extension of restructuring to Tagalog, there
may be others that also explain the data (see for instance Chung (1990)).
4.3. Agreement and Person Effects
We have shown above that a structure for modal sentences in which the subject appears below the
lower verb but Agrees with the modal is possible in Tagalog. The theory developed in Richards
(2005) allows us to confirm that the subject is indeed Agreeing with the modal in the transparent
clause structure. Richards argues that when a Probe Agrees with more than one Goal, Goals after
the first one must be in third person.
For example, Tagalog allows ANG-marked subjects to be fronted and marked with ay (i.e. ayfronted):5
5

Examples (16) and (17) are taken from Richards (2005).

8
(16)

UCLA Working Papers in Linguistics, no. 12

a.

Pilipino ako.
Filipino ANG-I
I am Filipino.

b.

Ako ay Pilipino.
ANG-I AY Filipino
I am Filipino.

However, a subject can be ay-fronted across a clause boundary only if it is in third person.
(17)

a.

Ang guro ay sinabi

ng mga pulis [na nagnakaw ti ng kotse.]
ANG teacher AY O.A.-said NG PL police [LI S.A.-stole ti NG car]
The teacher, the police said ti stole a car.

b.

Siya
ay sinabi
ng mga pulis [na nagnakaw ti ng kotse.]
ANG-he/she AY O.A.-said NG PL police [LI S.A.-stole ti NG car]
He/she, the police said ti stole a car.

c.

Ako/ka
ay sinabi
ng mga pulis [na nagnakaw ti ng kotse.]
ANG-I/ANG-you AY O.A.-said NG PL police [LI S.A.-stole ti NG car]
I/you, the police said ti stole a car.

Richards (2005) shows that in the examples above the Probe v Agrees first with the lower clause
and then with the subject. The Agree relation between the Probe v and the subject is thus not the
first one that v enters into, and so the subject is required to be in third person.
We can make a similar prediction for the transparent clause structure. If the modal is Agreeing
with the subject, then it will first Agree with the lower clause, and then with the subject NP, which
is inside the lower clause. The subject in the transparent clause structure will thus be required
to be in third person. And indeed, while third person subjects are allowed in all four types of
constructions discussed, first person and second person subjects are prohibited in transparent clause
constructions.
(18)

First/second person subject:

a.

Kailangan ko/mo
[bumili ng kotse.]
must
NG-I/NG-you S.A.-buy NG car

(control)

Asarina & Holt, Syntax and Semantics of Tagalog Modals

I/you must buy a car.

b. *Kailangan [bumili ko/mo ng kotse.]

(19)

(transparent clause)

c.

Kailangan [bumili ako/ka

ng kotse.]
must
S.A.-buy ANG-I/ANG-you NG car

d.

Kailangan ako/ka [bumili ng kotse.]

(opaque clause)

(raising)

Third person subject:

a.

Kailangan niya
[bumili ng kotse.]
must
NG-he/she S.A.-buy NG car
He/she must buy a car.

b.

Kailangan [bumili niya ng kotse.]

c.

Kailangan [bumili siya

ng kotse.]
must
S.A.-buy ANG-he/she NG car

d.

Kailangan siya [bumili ng kotse.]

(control)

(transparent clause)
(opaque clause)

(raising)

The modal thus Agrees first with the lower clause and then with the subject in the transparent
clause structure. In the control structure the first Agree relation that the modal enters into is with
the subject, as the is generated in the higher clause, and so first and second person subjects are
permitted. The subject in opaque clause and raising structures Agrees only with the lower verb, so
that again a subject in any person is allowed.

4.4. Opaque Clause

The third possibility to consider is that the subject appears after the lower verb, and is ANG-marked
by the lower verb, as in (7a) (repeated below).
(20)

Kailangan [ CP bumili ang lalaki ng kotse.]

[ CP S.A.-buy ANG man NG car]
must
The man must buy a car.

10

UCLA Working Papers in Linguistics, no. 12

There is room for negation to attach to the lower clause, as shown in (19a), so the lower clause
is certainly larger than a VP. Lacking counter-evidence, we suppose that the lower clause is just a
full CP. The modal is thus blocked from marking the subject and assigning a -role to it, and the
subject is instead marked by the lower verb.
4.5. Raising
Finally, it is also possible for the subject to be ANG-marked, as specified by the lower verb, but
to appear between the modal and the lower verb, as in (7b). As in the control structure (see (10)),
in the raising structure a linker can be inserted between the subject and the lower verb, which
indicates that the subject is in the higher clause.
(21)

Kailangan ang lalakii [ CP (na) bumili ti ng kotse.]

must
ANG mani [ CP (LI) S.A.-buy ti NG car]
The man must buy a car.

In order to be marked by the lower verb, the subject must have been generated in the lower clause
and raised to its surface position. There is no -role transmission from the modal to the subject,
even though the subject appears high in the sentence. This is, of course, consistent with standard
examples of raising: the raised subject receives a -role from the lower verb rather than the higher
one, as in (22).
(22)

The mani seems [ IP ti to be buying a car.]

5. S EMANTIC I MPLICATIONS

OF

S YNTACTIC S TRUCTURES

FOR

Kailangan

It has been stated above that a Tagalog modal assigns a -role to the subject if, and only if, it marks
the subject with NG. This means that when the subject in a sentence containing kailangan is NGmarked despite the presence of a S.A. lower verb, the reading of the sentence is directed deontic.
The obligation is assigned to the subject, and therefore the subject must be pragmatically able to
receive obligation. Conversely, if the subject is ANG-marked, the semantics of the lower clause
determines where obligation should be assigned, and the reading is non-directed deontic.
Consider the following example:
(23)

a.

The man must disappear.

Asarina & Holt, Syntax and Semantics of Tagalog Modals

b.

11

The bread must disappear.

In English, (23a) is ambiguous: it is possible that the man is obligated to disappear (directed
deontic), or that another contextually salient force has to make the man disappear (non-directed
deontic). In (23b), it is assumed for pragmatic reasons that the bread itself is under no obligation
to do anythingsome other contextually salient entity is obligated to make the bread go away. The
reading for (23b) thus has to be non-directed deontic. In Tagalog, the non-directed deontic reading
is available when the subject is marked by the lower verb, i.e. the structure is opaque or raising.
(Note that the intransitive verb mawala (disappear) below normally ANG-marks the subject.)
(24)

a.

Kailangan mawala ang lalaki/tinapay.

must
disappear ANG man/bread
The man/bread must disappear.

(opaque clause)

b.

Kailangan ang lalaki/tinapay mawala.

must
ANG man/bread disappear
The man/bread must disappear.

(raising)

However, only a directed deontic reading is allowed when the subject is NG-marked by kailangan.
This means that lalaki (man) can be the subject in a control or transparent clause sentence, as a
man can receive obligation. Tinapay (bread), on the other hand, cannot be the subject of a control
or transparent clause sentence, as it is pragmatically unacceptable for bread to receive obligation.
(25)

(26)

a.

Kailangan ng lalaki mawala.

(control)
must
NG man disappear
The man must make himself disappear.

b.

Kailangan mawala ng lalaki. (transparent clause)

must
disappear NG man
The man must make himself disappear.

a. #Kailangan ng tinapay mawala.

(control)
must
NG bread disappear
#The bread must make itself disappear.
b. #Kailangan mawala ng tinapay. (transparent clause)
must
disappear NG bread
#The bread must make itself disappear.

12

UCLA Working Papers in Linguistics, no. 12

6. S EMANTIC I MPLICATIONS

OF

S YNTACTIC S TRUCTURES

FOR

OTHER TAGALOG M ODALS

Kailangan lends itself equally well to structures in which the modal assigns obligation and a NGmarking to the subject and structures in which it does not. This is not the case for other Tagalog
modals. However, if a modal is able to NG-mark the subject and assign the -role to the subject,
then it allows both the control and transparent clause structures. Similarly, if the modal is able
to not NG-mark the subject, it allows both the opaque clause and raising structures. The key
distinction thus lies between modals that assign a NG marking and -role to the subject and those
that do not.
As has been previously mentioned, a directed deontic reading is obtained when the modal NGmarks the subject, and a non-directed deontic reading is obtained when the modal does not NGmark the subject. The dynamic reading is preferred with a modal that NG-marks the subject, but
is permitted for modals that do not.
We now discuss each of the remaining Tagalog modals.
6.1. Dapat
Dapat, which like kailangan can be translated as must, demonstrates a clear preference for structures in which the modal does not mark the subject (opaque clause and raising). The structures
in which the modal does mark the subject (control and transparent clause) are allowed for some
speakers, though they are dispreferred. The semantic facts for dapat are consistent with those for
kailangan: dapat assigns obligation to the subject if, and only if, it NG-marks the subject. 6
6.2. Kaya

Kaya, which may be translated as can, always NG-marks the subject.

6

There are semantic differences between kailangan and dapat even when both take a non-directed reading. For
example, (ia) is good, while (ib) requires a special context.
(i)

a.

b.

Kailangan umulan bukas

para ako
gumawa ng mainit na salabat.
must
rain
tomorrow for ANG-I make LI hot
LI ginger tea
It must rain tomorrow for me to make hot ginger tea.
#Dapat umulan bukas
para ako
gumawa ng mainit na salabat.
must rain
tomorrow for ANG-I make LI hot
LI ginger tea
It must rain tomorrow for me to make hot ginger tea.

It appears that kailangan sentences are consistent with the rules of the universe, while dapat sentences are consistent
with speaker authority. Hence, in (ib) it is odd to use dapat, as the speaker has no control of the weather. (ib) becomes
good, however, in the special context in which the speaker is a wizard who can control the weather through magic.

Asarina & Holt, Syntax and Semantics of Tagalog Modals

(27)

(28)

a.

Kaya ng lalaki bumili ng kotse.

can NG man S.A.-buy NG car
The man can buy a car.

b.

Kaya bumili ng lalaki ng kotse.

(control)

(transparent clause)

a. *Kaya bumili ang lalaki ng kotse.

can S.A.-buy ANG man NG car
b. *Kaya ang lalaki bumili ng kotse.

13

(opaque clause)

(raising)

Semantically, kaya always takes the dynamic reading. Since kaya NG-marks the subject, it always assigns ability to the subject in particular, as predicted by Brennan (1993). Subjects that
pragmatically cannot have abilities, such as bread, therefore cannot be used with kaya.
(29)

(30)

a.

Kaya ng lalaki mawala.

(control)
can NG man disappear
The man is able to disappear.

b.

Kaya mawala ng lalaki. (transparent clause)

can disappear NG man
The man is able to disappear.

a. #Kaya ng tinapay mawala.

(control)
can NG bread disappear
#The bread is able to disappear.
b. #Kaya mawala ng tinapay. (transparent clause)
can disappear NG bread
#The bread is able to disappear.
6.3. Puwede and Maaari

Puwede and maaari may both be translated as can. There seems to be little semantic difference
between the two.7 When the lower verb is S.A., both puwede and maaari require an ANG-marked
7

Puwede is a Spanish borrowing and is more commonly used in conversation.

14

UCLA Working Papers in Linguistics, no. 12

subject.
(31)

a. *Puwede/Maaari ng lalaki bumili ng kotse.

can
NG man S.A.-buy NG car
b. *Puwede/Maaari bumili ng lalaki ng kotse.

(32)

(control)

(transparent clause)

a.

Puwede/Maaari bumili ang lalaki ng kotse.

can
S.A.-buy ANG man NG car
The man can buy a car.

b.

Puwede/Maaari ang lalaki bumili ng kotse.

(opaque clause)

(raising)

Correspondingly, since the subject always receives its marking from the lower verb, puwede and
maaari never assign a -role to the subject. The two modals thus take a non-directed deontic
reading, never a directed one. Consequently, subjects that pragmatically cannot receive permission
are allowed in puwede and maaari sentences, both in opaque clause and raising structures.
(33)

(34)

a.

Puwedeng/Maaaring mawala ang lalaki.

can-LI
disappear ANG man
It is permissible for the man to disappear.

b.

Puwedeng/Maaaring mawala ang tinapay.

can-LI
disappear ANG bread
It is permissible for the bread to disappear.

a.

Maaaring/Puwedeng ang lalaki mawala.

can-LI
ANG man disappear
It is permissible for the man to disappear.

b.

Puwedeng/Maaaring ang tinapay mawala.

can-LI
ANG bread disappear
It is permissible for the bread to disappear.

(opaque clause)

(opaque clause)

(raising)

(raising)

Though kaya is strongly preferred in dynamic contexts, a dynamic reading is also allowed for
puwede and maaari, This appears to contradict Brennans claim that dynamic modals always take
the subject as an argument and assign a -role to it. Of course, Brennan was making arguments
about English rather than Tagalog, but it is reasonable to suppose that there is consistency between
languages, as such consistency appears to exist for deontic modality. It is possible, however, that

Asarina & Holt, Syntax and Semantics of Tagalog Modals

15

puwede and maaari do not get real dynamic readings. Able to means approximately permitted
to by certain facts, and perhaps puwede and maaari are taking this particular permission reading
when they take an ability reading. This is consistent with the two modals being dispreferred for
situations when the dynamic reading is required.
Another possibility is that puwede and maaari take a non-directed dynamic reading, in which
ability is not assigned to the subject in particular. However, there is a lack of separate evidence for
such a division between dynamic modals, and it is unclear what it would mean for ability not to
be assigned directly to the subject. This approach therefore seems less desirable than the first, and
also does not explain the preference for using kaya, rather than puwede and maaari, in dynamic
contexts.
7. S UMMARY OF FACTS
Marking on Subject
Structures
Readings
Modals

From Modal
control, transparent clause
directed deontic, dynamic

From Lower Verb

opaque clause, raising
non-directed deontic, dynamic
(dispreferred)
kailangan, kaya, dapat (dispreferred) kailangan, dapat, puwede, maaari
8. C ONCLUSION

Tagalog provides clear evidence that semantic differences between modals correspond to syntactic
differences. In particular, non-directed deontic modals take a CP argument and do not assign a role to the subject. Directed deontic modals take the subject NP and a clause, and assign a -role to
the subject. In Tagalog, non-directed deontic modals do not syntactically mark the subject, while
directed deontic modals assign a NG-marking to the subject. The dynamic reading is preferred
with a modal that assigns a NG-marking and a -role to the subject, though it seems to also be
allowed for modals that do not. The Tagalog evidence is thus basically in line with the proposal
made for English modals by Brennan (1993).
Appendix I. Pronominal Variable Binding
We can use binding theory to attempt to determine how high the subject is in raising and control constructions. Rackowski (2002) shows that an ANG-marked object can bind a pronominal
variable in the subject:
(35)

Minamahal ng kanyang
ama ang bawat anak.
O.A.-love NG POSSESSIVE father ANG every child
Hisi /heri father loves every childi.

16

UCLA Working Papers in Linguistics, no. 12

An ANG-marked object must then also be able to bind a variable in the subject in a raising construction, since raising should not affect binding. We know that puwede and maaari allow raising
and disallow control constructions, so we can use these modals to test the prediction. And indeed,
an ANG-marked object can bind a variable in the subject in raising structures with puwede and
maaari.8
(36)

Puwede/maaari ng kanyang ama (na) mahalin

ang bawat anak.
can
NG possessive father (LI) O.A.-love-INF ANG every child
Hisi /heri father can love every childi .

Surprisingly, however, the ANG-marked object is allowed to bind a variable in the subject in
control structures as well.
(37)

Kaya ng kanyang ama (na) mahalin

ang bawat anak.
can NG possessive father (LI) O.A.-love-INF ANG every child
Hisi /heri father can love every childi .

Kaya allows control structures, but not raising structures, so we would expect that the subject in
(37) appears too high for the object to scope over it and bind a variable. Further research is needed
to determine why the object can scope over the subject in (37).
References
B OBALJIK , J ONATHAN and S USI W URMBRAND. 1999. Modals, Raising and A-Reconstruction.
Handout for talk given at Leiden University (October 1999), University of Salzburg (December
1999).
B RENNAN , V IRGINIA. 1993. Root and Epistemic Modal Auxiliary Verbs. Ph.D. thesis, UMass,
Amherst.
C HUNG , S ANDRA. 1990. VPs and Verb Movement in Chamorro. Natural Language and Linguistic Theory 8, 559619.
H ALL , DANIEL. 2001. The Featural Semantics of English Modal Verbs. Generals Paper in Semantics, University of Toronto.
JACKENDOFF , R AY S. 1972. Semantic Interpretation in Generative Grammar. Cambridge, MA:
MIT Press.
M ERCADO , R APHAEL. 2002. Verb Raising and Restructuring in Tagalog. MA forum paper, University of Toronto.
. 2005. XP Predicate-Fronting in Tagalog. Talk given at AFLA XII.
R ACKOWSKI , A NDREA. 2002. The Structure of Tagalog: Specificity, Voice, and the Distribution
of Arguments. Ph.D. thesis, MIT.
8

These sentences are difficult to parse, and some context is required to make them acceptable.

Asarina & Holt, Syntax and Semantics of Tagalog Modals

17

R ICHARDS , N ORVIN. 2005. Person-Case Effects in Tagalog and the Nature of Long-Distance
Extraction. In Proceedings of AFLA XII.
R IZZI , L UIGI. 1982. Issues in Italian Syntax. Dordrecht: Foris.
W URMBRAND , S USI. 2001. Infinitives: Restructuring and Clause Structure. Berlin/New York:
Mouton de Gruyter.
. 2002. Syntactic vs Semantic Control. In Studies in Comparative Germanic Syntax
(Z WART, JAN -WOUTER and W ERNER A BRAHAM, eds.), pp. 95129, Amsterdam: John Benjamins.