Transcript ch6

Chapter six
The syntactic derivation of Double
object construction in Arabic
• the dative sentence has a DO and an IO, and
that
• the IO in Arabic is preceded by the preposition
/i 'to' as appears in the examples, of S
• initial structures in (1) :
• )1a) zayd-un ?a9Taa kitaab-an li-hind-in
• Zayd-nom gave book-acc to-Hind-gen
• 'Zayd gave a book to Hind'
•
•
•
•
•
•
•
•
•
•
•
•
b? zayd-un ?a9Taa li-hind-in kitaab-an
Zayd-nom gave to-Hind-gen book-ace
c* zayd-un ?a9Taa li-hind-in
Zayd-nom gave to-Hind-gen
d?? zayd-un ?a9taa kitaab-an
Zayd-nom gave book-ace
e zayd-un ?a9taa hind-an kitaab-an
Zayd-nom gave Hind-ace book-ace
'Zayd gave Hind a book'
• Based on (1), the general properties of datives
can be immediately established.
• First, the dative construction exhibits a DO + pp
complement structure as appears in
• the well-formed (1a). Secondly, sentences with
the alternative PP+DO structure. are
• not fully accepted (1b). Third, the ill-formed
sentence in (1c) shows that the deletion of the
DO is not tolerated and (44d) is marginal due to
the absence of the PO.
• The well-formed sentence in (1e) represents the
DOC where the IO precedes the DO. .
• To account for the derivation of datives and
DOCs, I will suggest that the
• former is derived from the latter. This proposal
requires two assumptions. The first is
• that the IO is a PP in all positions, and that the
prepositional head of the PP is null in
• DOCs, i.e., is not realized phonologically, if and
only if the PP is governed by a Case assigning
verb. Second, the derivation of datives relies on
the lexical preposition preceding the IO and on
Larson's notions of V' -reanalysis and complex
predicate' as will be illustrated in Section 5.5 .To
clarify how the DOC is derived, we first assume
the partial D-structure in (2)Link
• This is incompatible with the word order of
the DOC unless we assume an empty verb
position to the left of the pp (IO) at Dstructure; then we can derive the S- initial
word order by movement. I therefore propose
(3) below as the D-structure representation of
DOCs and datives, and assume that the
surface word order of DOCs is derived by the
movement of the verb to a position to the left
of the IO which is base generated as the head
of a higher VP.
• Assuming also that the subject is base
generated in the specifier of the higher VP (c!
Kitagawa, 1986; Kuroda, 1988; Koopman and
• Sportiche, 1988), (3) yields (4) following Verb
raising (ultimately to I) and subject
• movement:
• Link
• The D-structure of DOCs in (4 ) can be
motivated in various ways. First, the Theme is
realized as an 'inner' DO lower in the tree.
Plausibility for this view can be .derived from
the fact that this object has an intuitively
'closer' semantic and syntactic relation to the
verb than does the IO in both DOCs and
datives, as is indicated by the observation that
the latter can be omitted in some cases (due,
we suppose, to a lexical property of specific
verbs), but not the former.
• By way of illustration consider the
• sentences in (5) and (6) below:
• 5a hal baa9-a zayd-un hind-an kitaab-an?
• Q sold Zayd-nom Hind-ace book-ace
• 'DtdZayd sell Hind a book'?
• b hal baa9-a zayd-un kitaab-an?
• Q sold Zayd-nom book-ace
• 'Did Zayd sell a book'?
• c* hal baa9-a zayd-un hind-an?
• Q sold Zayd-nom Hind-ace
• 6a hal arsal-a zayd-un risaala-tan li-hind-in
• Q sent Zayd-nom letter-ace to-Hind-gen
• 'Did Zayd send a letter to Hind'?
• b hal arsal-a zayd-un risaala-tan
• Q sent Zayd-nom letter-ace
• 'Did Zayd send a letter'?
• c* hal arsal-a zayd-un li-hind-in
• Q sent Zayd-nom to-Hind-gen
• Due to the occurrence of the two objects, (5a)
and (6a) are grammatical. In 5b) and (6b) the
sentences are grammatical even though the IO is
omitted, whereas 5c) and (6c) are deviant
because of the deletion of the DO.
• Second, (5) clearly involves a claim that the IO as
a pp appears in specifier position and the DO
appears as a complement. In this section we shall
see how this analysis enables us to produce a
straightforward account of how DOCs work.
• Third, we assume that although there are two
sorts of Case (structural and
• inherent), these Cases are assigned in the same
configuration. In (5), we have a situation of a
single Case assigner and two arguments which
need Case. These arguments are in different
positions, therefore they cannot both be in the
right configuration. Consequently, it must be the
case that the verb can move so' that it appears in
two different configurations, each of which is
appropriate for one of the arguments and it is the
empty verb position that creates this possibility .
• The IF contains another head position which
allows the verb to move in a further step to
get tense and agreement from INFL which
assigns Nominative Case to the subject under
spec-head agreement. The subject moves
from its base position to the higher spec of IP
to yield the S-initial word order and to be
assigned Nominative Case under spec-head
agreement.
• Case assignment
• After outlining the derivation of the DOC, we
move to investigate precisely how Case is
assigned to the two objects in this
construction.
• As noted, the S-structure of the DOC in (6)
poses a problem for Case theory in that there
are two NPs which must receive Case in order
to pass the Case Filter. We suppose that verbs
in MSA and Palestinian Arabic (PA), however,
can as in most languages only assignstructural
Case to one NP
• The obvious question is: what about the other
NP? The issue raised is of course identical to
that of how the second NP in an English DOC
like (7) is assigned Case:
• 7. John gave Mary a book
• According to the proposal of Chomsky (1980),
some verbs can assign another type of Case,
Inherent Case, in addition to structural Case.
• Extending this idea to Arabic ditransitive verbs
will provide them with enough Case assigning
potential to ensure that their arguments
satisfy the Case Filter .Next, we have to
consider the issue of which object receives the
structural Case, and which object has the
Inherent Case in the DOC and why. Before
investigating this matter, it is crucial to note
that Inherent Case is assumed to differ from
structural Case in one very important respect.
• We suppose that the former is assigned under
government at D-structure, and the assigning
head must theta mark the relevant NP. By
contrast, the latter is assigned under
government at S-structure, and there need
not be any direct thematic relationship
between the assigning head and the NP.
• Modifying this, we might suppose that
structural Case can be assigned at S structure
or at intermediate levels in a derivation. We
can then suggest that the verb, in its base
generated position assigns Inherent
Accusative Case to the DO at D-structure.Then
it raises to the empty verb position, and
discharges its structural Case in the empty
verb position to the IO via the empty
preposition.
• Finally, it raises to I to be inf1ected and then,
following Koopman and Sportiche (1991), Ouhalla
(1991), among others, the external argument is
assigned structural Nominative Case from I under
spec-head agreement.
• The DO which is always base generated in the
lower complement position in ditransitive clauses
of MSA and PA cannot be promoted under
passivization
• If we suppose that Inherent Case is retained
under grammatical processes, we now have an
account of this asymmetry.
• Given this analysis, Case assignment to the
subject and the two objects in (8 a) can be
structurally represented as in ( 8 b)
•
• 8a)zayd-un ?a9Taa hind-an kitaab-an
• Zayd-nom gave Hind-acc book-acc
• 'Zayd gave Hind a book'
• Link
• This schema indicates clearly how the analysis is
consistent with some common assumptions
about Case assignment. First, the verb's structural
Case is assigned to the most adjacent' object
hind, where 'adjacency' is computed during the
derivation or at S-structure. This leaves only
Inherent Case available which is assigned to the
argument of the verb kitaab at D-structure.
• Second, the structurally Case marked
• intervenes between the Inherently Case marked
NP and the verb.
• Having formulated a proposal as to how
arguments are assigned Case in DOCs, we move
next to consider theta role assignment
• Theta -role assignment
• Ditransitive verbs have three theta-roles to
assign. In this section we shall consider how this
process occurs. In pursuing this, echoing to some
extent Falk (1990). we shall assume a theta
theory
• based on (9)
• 9. Theme: assigned directly by the verb
• Possessor: indirectly assigned via a higher
• projection of the verb.
• Goal: assigned directly by a governing
preposition
• Agent: assigned compositionally by verb + Theme
+ Possessor (or Goal)
• We shall first see how (9) works in a completely
mechanical fashion. Then we shall look for some
evidence for it.
• According to the proposed theory, and in line
with Falk (1990), the verb in the lower position
directly assigns Theme to the DO which is base
generated in the complement position and is
canonically governed by this verb. Diverging from
Falk's proposal, the PO which is base generated
as part of the PP in [spec, VP] is assigned
Possessor theta-role compositionally via a higher
projection (V') of the lower verb.
• This theta role is directly assigned via the next
higher projection (the lower V') under
sisterhood to the PP and then it is transmitted
via the empty preposition, which is not a
theta-role assigner, to the IO. The process of
transmission through the null preposition
entails that this theta-role is assigned
indirectly
• Theta-role assignment in datives
• According to (9) above, in datives, the DO is assigned
Theme theta-role directly by the verb at D-structure,
whereas the PO is assigned Goal theta-role by the
lexical preposition preceding it; unlike the null
preposition, the lexical preposition has an inherent
theta-role to assign, and the question of having the
theta-role assigned compositionally does not arise. We
therefore maintain that, although the DOC and
• dative construction have the same syntactic
configuration at D-structure, the choice of lexical
versus empty preposition actually triggers a different
mode of theta-role assignment in the two cases; the
theta-role of the complement of the lexical PP must be
licensed by a strategy different from that licensing the
IO in DOCs above and we assume this to be the dative
preposition li..
• Datives in Hebrew
• Hebrew offers no motivation for a productive
relationship between DOCs and dative constructions.
According to Givon (1984)
• there is no dative shifting via which an indirect
(prepositional, object (IO) may lose its semantic Case.
Accordingly, only the DO can appear as a bare
accusative (cf also Belletti and Shlonsky, 1995).
Consider (10) and (11) :
• .10a . Zayd natan sefer la-hind
• Zayd gave book to-Hind
• 'Zayd gave a book to Hind'
• b zayd natan la-hind sefer
• Zayd gave to-Hind book
• 'Zayd gave to Hind a book'
• * zayd natan hind sefer
• Zayd gave Hind book'
• 11a. ha mone hesbiir
it ha-oi9uur la talmiid
• The teacher explained acc the-lesson to-the-pupil
• b ha mone hesbiir la talmiid it ha-oi9uur
• The teacher explained to the pupil acc the-lesson
• c* ha mone hesbiir it ha-oi9uur talmiid
• The teacher explained acc the lesson the pupil
• As these examples show, Hebrew,. unlike Arabic,
does not accept the DOC, and this raises the
question of why this language does not accept
this construction while Arabic does .This question
has been answered in a variety of ways in the
literature. Larson (1988) connected the
availability of DOCs with P-stranding. His
generalisation, following Kayne (1984), is that
languages which accept dative shift also accept Pstanding, and not vice versa. As Hebrew does not
have either DOCs or P-stranding, it is consistent
with this generalisation.
• However, as we have seen, the generalisation is directly
contradicted by Arabic which in spite of fallowing
dative shift does not accept P-stranding. Obviously a
generalisation which is so blatantly falsified cannot
form the basis for an explanation.
•
• Another attempt to deal with the same phenomena
appears in Tremblay (1990).
• He claims that the possibility of having dative shift is
directly related to the possibility of having head-final
NPs [NP N] languages which have head-final NPs accept
dative shift while languages which do not have headfinal NPs do not accept dative shift .Illustrative
examples from English and French are from Tremblay
(1990: 552)
•
•
•
•
•
•
•
12a Jean gave Mary a book
b Mary's book
13a * J eanne a donne Marie un livre
b* Mane livre
Again, this correlation is confounded by Arabic
and so can hardly be used to explain the absence
of DOCs in Hebrew. Although the two Semitic
Languages have head initial NPs, Arabic allows
DOCs while Hebrew does not. Possessive NPs in
Arabic and Hebrew are exemplified in (14) and
(15)
•
•
•
•
•
•
•
14 kitaab-u hind-in
book-nom Hind-gen
'Hind's book'
15 sefer ha-saxkan
Book the actor
'The actor's book'
• On the basis of the above, it is necessary to
find another strategy to account for the
presence of DOCs in Arabic and English and
their absence in Hebrew and other
languages .Patterning to the account
developed in this chapter, we might suggest
that
• Hebrew, French and other languages lack the
option of an empty preposition strategy for
syntactically realising a Possessor argument.
• In other words, having or not having an empty
preposition strategy is entirely equivalent to
having or not having a DOC in a language. To
the extent that this is plausible, it has the
consequence that the Hebrew
• verb natan lacks the full semantic potential of
English give and Arabic ?a9Taa.
• Dative and Double object constructions in
English
• Regarding the dative alternation, English has
three categories of verbs like those of Arabic
investigated above. This immediately entails the
conclusion that the analysis developed for Arabic
above can be applied to English without
significant modification. To remind the reader,
many verbs
• display a productive relationship between DOCs
and dative constructions. Ditransitive
• verbs generally have alternate forms with the IO
in a pp as shown in (16-17),
•
•
•
•
16a She gave him a book
b* She gave to him a book
c She gave a book to him
d * She gave a book him
•
•
•
•
17a John threw Mary the ball
b* John threw to Mary the ball
c John threw the ball to Mary
d* John threw the ball Mary
•
•
•
•
18a He paid her one pound
b* He paid to her one pound
c He paid one pound to her
d* He paid one pound her
• As can be seen, the structure of the sentences
above are identical in the relevant. respects to
their counterparts in Arabic, and this yields a
straightforward application of the analysis
developed in this chapter.
• However, the memberships of the three
categories of verbs are not identical across the
two languages, and it is necessary to address
these differences before concluding this
chapter.
• Semantic constraints
• It has been claimed that the range of verbs
that participate in the DOC is relatively narrow
in Arabic, whereas English has a wide range of
verbs which appear in this construction. Thus,
in comparing the English verbs which
participate in DOCs with their near synonyms
in Arabic, we find a lack of correspondence
across the two languages. For convenience,
consider the English and Arabic verbs listed in
(5) (6) and (7) below
•
•
•
•
•
•
•
•
•
•
•
•
•
List 5): alternating verbs in English and Arabic
Alternating verbs
English
Arabic
givepass
?a9Taa 'gave'
pay post
?9aar-a 'borrowed'
kick feed? saIl am-a 'handed'
trade?
e-mail wahab-a 'granted'
promtse
hand
baa9-a 'sold'
Telephone buy nawal-a 'handed'
throw
get
manaH-a 'granted'
flick bring ?qraD-a 'borrowed'
lend
radio
?hdaa 'gifted'
grant
offer
wa9ad-a 'promise'
•
•
•
•
•
•
assIgn sell
WIre
serve
Teach satellite
tell send
toss
make
loan
telegraph
•
•
•
•
•
•
•
•
•
•
•
•
•
•
)
verbs participating in only DOCs in English and Arabic
Verbs allowing only DOCs
English
Arabic
cost kallaf-a 'cost'
ask
sa?a/-a 'asked'
bet
kasaa 'bought clothes for someone
save ? axbar-a 'told'
deny
razaq-a'sustained'
charge kafa?-a'rewarded'
refuse
da9aa 'named'
spare
kanaa 'named'
fine waqaa 'avoided'
forgive
•
•
•
•
•
•
•
•
•
•
•
•
•
•
7)
verbs participating in only datives in English and Arabic
Verbs accepting only datives
English verbs
Arabic verbs
donate
&rraH-a 'explained'
contribute ?r&rd-a 'guided'
distribute qaddam-a 'offered'
say
katab-a 'wrote'
push
?rsa/-a 'sent'
carry
?aHDar-a 'brought'
report
wajjah-a'directed'
pull
ram a 'kicked'
lift
naqa/-a 'carried'
ease
DabaH-a'slaughtered'
•
•
•
•
•
•
?abraq-a'telegramed‘
tabara9-a 'donated‘
?a9aad-a'returned‘
zawwaj-a 'marry a female to male‘
xaTab-a 'have a female engaged to male'
• The lack of correspondence between the verbs
appearing in the tables above gives rise to the
question of how is the variation between the two
languages to be accounted for?
•
• Regarding this question, we propose that the variation
between the two
• languages in the number and identity of verbs which
either alternate or do not hinge on rather subtle
semantic issues. Both languages have the null
preposition option, so the differences cannot be due to
the major syntactic choice. We propose, then, that
some verbs allow the options of both nulll1exical
preposition (the alternating verbs). and others do not.
This, in turn, comes down to the lexical entry of verbs,
with some verbs
• allowing only the Goal or Possessor theta-role in one or
other language. That is, there are relatively slight
differences in the meaning potentials of cognate verbs
in the two
• languages, a not unexpected conclusion in the
light of cross-linguistic investigation. of semantic
fields. This possibility for variation between
English and Arabic in the number of verbs which
alternate, could, in principle, be investigated in
terms of a more structured set of semantic
classes These may include: possessional verbs
whose Goal is an animate (e.g., give), animate
control verbs (e.g., pass), verbs with an
informational
• dimension with an animate Goal (e.g., tell), and
positional verbs such as throw (Gruber, 1992,
Lefebvre, 1994).
• Following Lefebvre's account of Fongbe in
spirit, the counterpart verbs in Arabic might
be limited to. the possessional verbs (e.g.,
?a9Taa) and verbs with an informational
dimension, (e.g., wa9ad-a) and this might
account for the limited number of verbs which
either alternate or only accept DOCs in Arabic.
•