Chapter six The syntactic derivation of Double object

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) : • (1 a) zayd-un ? a 9 Taa kitaab-an li-hind-in • Zayd-nom gave book-acc to-Hind-gen • 'Zayd gave a book to Hind'

• • • b? zayd-un ? a 9 Taa li-hind-in kitaab-an Zayd-nom gave to-Hind-gen book-ace c* zayd-un ? a 9 Taa li-hind-in Zayd-nom gave to-Hind-gen d? ? zayd-un ? a 9 taa kitaab-an Zayd-nom gave book-ace e zayd-un ? a 9 taa 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 (1 a). Secondly, sentences with the alternative PP+DO structure. are • not fully accepted (1 b). Third, the ill-formed sentence in (1 c) shows that the deletion of the DO is not tolerated and (44 d) is marginal due to the absence of the PO. • The well-formed sentence in (1 e) 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: • 5 a hal baa 9 -a zayd-un hind-an kitaab-an? • Q sold Zayd-nom Hind-ace book-ace • 'Dtd. Zayd sell Hind a book'? • b hal baa 9 -a zayd-un kitaab-an? • Q sold Zayd-nom book-ace • 'Did Zayd sell a book'?

• c* hal baa 9 -a zayd-un hind-an? • Q sold Zayd-nom Hind-ace • 6 a 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, (5 a) and (6 a) are grammatical. In 5 b) and (6 b) the sentences are grammatical even though the IO is omitted, whereas 5 c) and (6 c) 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 inf 1 ected 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) • • 8 a(zayd-un ? a 9 Taa 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 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; 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) : • . 10 a. 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' • 11 a. ha mone hesbiir it ha-oi 9 uur la talmiid • The teacher explained acc the-lesson to-the-pupil • b ha mone hesbiir la talmiid it ha-oi 9 uur • The teacher explained to the pupil acc the-lesson • c* ha mone hesbiir it ha-oi 9 uur 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 head-final NPs do not accept dative shift. Illustrative examples from English and French are from Tremblay (1990: 552)

• • 12 a Jean gave Mary a book b Mary's book 13 a * 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 ? a 9 Taa.

• 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),

• • 16 a 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 • • 17 a 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 • • 18 a 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 ? a 9 Taa 'gave' pay post ? 9 aar-a 'borrowed' kick feed? sa. Il am-a 'handed' trade? e-mail wahab-a 'granted' promtse hand baa 9 -a 'sold' Telephone buy nawal-a 'handed' throw get mana. H-a 'granted' flick bring ? qra. D-a 'borrowed' lend radio ? hdaa 'gifted' grant offer wa 9 ad-a 'promise'

• • • ass. Ign 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 da 9 aa '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 &rra. H-a 'explained' contribute ? r&rd-a 'guided' distribute qaddam-a 'offered' say katab-a 'wrote' push ? rsa/-a 'sent' carry ? a. HDar-a 'brought' report wajjah-a'directed' pull ram a 'kicked' lift naqa/-a 'carried' ease Daba. H-a'slaughtered'

• • • ? abraq-a'telegramed‘ tabara 9 -a 'donated‘ ? a 9 aad-a'returned‘ zawwaj-a 'marry a female to male‘ xa. Tab-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 nulll 1 exical 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. , ? a 9 Taa) and verbs with an informational dimension, (e. g. , wa 9 ad-a) and this might account for the limited number of verbs which either alternate or only accept DOCs in Arabic. •