15
Full Terms & Conditions of access and use can be found at http://www.tandfonline.com/action/journalInformation?journalCode=rwrd20 Download by: [104.157.129.119] Date: 29 July 2016, At: 17:13 <i>WORD</i> ISSN: 0043-7956 (Print) 2373-5112 (Online) Journal homepage: http://www.tandfonline.com/loi/rwrd20 PRO theory Ala Al-Kajela To cite this article: Ala Al-Kajela (2015) PRO theory, <i>WORD</i>, 61:1, 25-38, DOI: 10.1080/00437956.2015.1006855 To link to this article: http://dx.doi.org/10.1080/00437956.2015.1006855 Published online: 05 Mar 2015. Submit your article to this journal Article views: 283 View related articles View Crossmark data

PRO theory

Embed Size (px)

Citation preview

Full Terms & Conditions of access and use can be found athttp://www.tandfonline.com/action/journalInformation?journalCode=rwrd20

Download by: [104.157.129.119] Date: 29 July 2016, At: 17:13

<i>WORD</i>

ISSN: 0043-7956 (Print) 2373-5112 (Online) Journal homepage: http://www.tandfonline.com/loi/rwrd20

PRO theory

Ala Al-Kajela

To cite this article: Ala Al-Kajela (2015) PRO theory, <i>WORD</i>, 61:1, 25-38, DOI:10.1080/00437956.2015.1006855

To link to this article: http://dx.doi.org/10.1080/00437956.2015.1006855

Published online: 05 Mar 2015.

Submit your article to this journal

Article views: 283

View related articles

View Crossmark data

PRO theory

Ala Al-Kajela*

Department of Linguistics and Languages, McMaster University, 1280 Main Street West, Hamilton,ON L8S 4K1, Canada

(Received 12 January 2014; accepted 7 June 2014)

Control theory has received a lot of attention from syntacticians and semanticists on apar such as Chomsky (1965), Rosenbaum (1967), Hornstein (1997, 1999), Boeckxand Hornstein (2003), Landau (2003), Culicover and Willinks (1986) and Culicoverand Jackendoff (2001) to name but some. We attempt to investigate the controversialphenomenon of control in Standard Arabic (SA). The paper sheds some light onStandard Arabic grammar and shows how Rosenbaum’s Minimal Distance Principle isviolated by the Arabic elaborate morphological system. We try to illustrate that controlbehaves differently depending on the language in question and that syntactic andsemantic analyses cannot be employed in a language that is structurally different fromEnglish. One of these distinct differences is that Arabic syntax employs a silentelement, viz., pro instead of the anaphoric PRO.

Keywords: PRO; obligatory control; non-obligatory control; Standard Arabic; minimaldistance principle; government and binding

1. Laying the background

The striking superficial similarity between constructions cross-linguistically has alwaysintrigued linguists to investigate and scrutinize such intriguing issues in the theory ofsyntax. A case in point is control theory which has received an ample amount of attentionfrom syntacticians over the last four decades. For example, the sentences in (1) and (2)share identical surface structure (wording) but they have different syntactic analysis ingenerative syntax.

(1) John seemed to understand the problem.(2) John tried to understand the problem.

At first glance, we could say that these structures are similar except for having a differentverb in the matrix clause, though both are intransitive. However, a closer look wouldreveal that in the first sentence (which Subject Raising and is not our focus here) John hasnothing to do with the matrix verb from a thematic point of view, because the verb seemdoes not assign any theta role. However, John is responsible for doing the understandingaction in the embedded clause, as the predicate understand has thematic role to assign. Inother words, John is the experiencer. In sentence (2), John is thematically related to thematrix verb try and the embedded verb understand; John is the agent of matrix verb and

*Email: [email protected]

WORD, 2015Vol. 61, No. 1, 25–38, http://dx.doi.org/10.1080/00437956.2015.1006855

© 2015 International Linguistic Association

Dow

nloa

ded

by [

104.

157.

129.

119]

at 1

7:13

29

July

201

6

the experiencer of the embedded predicate understand. This might look like thetacriterion violation but it will become clear when we talk about theta role assignment. Thiskind of sentences is referred to as Subject Control.

Now if we consider the sentences in (3) and (4), we again come across two similarstrings which differ in having a distinct transitive matrix verbs.

(3) John imagined his mother to have met his girlfriend.(4) John urged his mother to meet his girlfriend.

In (3), the postverbal argument has one role to play that is of agent of the embeddedclause. The matrix verb imagine assigns a thematic role of agent to its subject John andtheme to the clausal complement; such constructions are called Object Raising. Whereasin (4) the postverbal argument his mother plays double role, the first is the object of thematrix verb and the second is the agent of the embedded verb meet. Thus, the matrix verburge assigns: (1) agent, (2) theme and (3) proposition to the clausal complement. Theconstruction in (3) is usually referred to as Raising to Object or Object Raising and theone in (4) is termed Object Control. It is worth noting that there is a battery of diagnosticswhich can be used to distinguish these identical constructions. We are listing some ofthem below not to show how both constructions contrast, but to illustrate how controlpredicates behave in the first place.

The abbreviations used in glosses of data:Comp = complementizer, du = dual, f = feminine, IMP = imperfective, m = masculine,PERF = perfective, POSS = possessive, s = singular, 1 = first person, 2 = second person,3 = third person.

1.1. Idiomatic meaning

It is a straightforward test that can be applied to differentiate between raising and controlconstructions. Raising constructions are known to maintain idiomaticity, whereas controlconstruction obscures idiomaticity. This can be explained by the fact that the subject ofthe idiom must be close to the rest of it and this is true for raising but not for control asshown below:

(5) The cat is out of the bag.(6) The cat feels that she is out of the bag.(7) The cat seemed to be out of the bag. (Idiosyncratic meaning)(8) ? The cat tried to be out of the bag. (Literal meaning)

1.2. Thematic relations

As we have pointed out above that the raising predicate does not assign thematic role toits subject, this argument gets a thematic role from the embedded predicate. On the otherhand, the control predicate assigns a thematic role to its subject (agent) and theme andproposition in case of object control, and agent to its subject and proposition to theclausal complement in case of subject control.

26 A. Al-Kajela

Dow

nloa

ded

by [

104.

157.

129.

119]

at 1

7:13

29

July

201

6

1.3. Selectional restrictions

Control predicate requires its agent to be sentient whether transitive or intransitive,whereas the semantics of embedded predicate determines the well-formedness of theraising construction.

(9) The pencil seems to be wood.(10) #The pencil seems to comprehend the lesson.(11) #The pencil tried to be wood.(12) #The pencil tried to understand the problem.

1.4. Pleonastic or (expletive) subjects

Expletives are allowed in non-thematic positions, which is a characteristic feature ofraising constructions.

(13) It seemed to be snowing.(14) *It tried to be snowing.(15) There seems to be a man in the backyard.(16) *There tried to be a man in the backyard.

2. Early analysis of control

Chomsky’s ‘Standard Theory’, 1965, and its components offered the first technique toderive control constructions. The principal components of this theory consisted of:

(1) Phrase Structure Rules (subcategorization rules);(2) Lexicon Insertion Rules (semantic representation);(3) Transformational Rules.

Following Rosenbaum (1967), we will apply Chomsky’s transformational rules tosentence (4) and the derivation is given below to show the order in which these rulesapply.

(4a) John urged [NP his mother] [S[NP his mother]] meet his girlfriend Deep Structure(4b) John urged [NP his mother][S for [NP his mother] to meet his girlfriend]

Complementizer Insertion(4c) John urged [NP his mother] [S for to meet his girlfriend] Equi(4d) John urged [NP his mother] [S to meet his girlfriend] Complementizer Deletion

The Equi (the short form for equivalent NP deletion) is the transformation that deletes thesubject NP of the embedded clause when its counterpart is either object of the matrixclause as in (4) above or subject as in (2).

3. PRO status

“There is an intrinsic fascination in the study of properties of empty categories”(Chomsky 1981: 55). We mentioned earlier that in (2) John is the agent of matrix verband the experiencer of the embedded predicate understand and in (4) the postverbal

WORD 27

Dow

nloa

ded

by [

104.

157.

129.

119]

at 1

7:13

29

July

201

6

argument his mother is the object of the matrix verb and the agent of the embedded verbmeet. These are considered clear violation of ϴ-criterion: “Each argument bears oneand only one ϴ-role, and each ϴ-role is assigned to one and only argument” (Chomsky1981: 36). This means that a structure will be ruled out if it has more arguments thanrequired, or if it has less arguments than required which leaves ϴ-roles unassigned.According to Chomsky, ϴ-roles are assigned at the LF to satisfy the subcategorizationrequirements of lexical items through all the syntactic levels of derivation.

It is clear then that ϴ-criterion will consider (2) and (4) ill-formed sentences.Therefore, there must be an argument in the embedded clause to get the experiencer rolein the matrix clause in (2) and the agent role in the embedded clause in (4). This distinctargument will satisfy ϴ-criterion and the Extended Projection Principle (EPP). Chomsky(1981) calls this special empty category PRO. He claims that PRO is null and silentbecause it appears in a caseless position. It is a DP that appears in the position of aspecifier of non-finite TP.

What are the features of PRO? PRO is a base generated phonetically null DP and it isnot a residue of movement like NP-traces. Chomsky and Lasnik (1993) propose that PROhas a ‘null’ Case and it is the only argument that bears this special Case. In fact, PRO hasthe feature [+anaphor, +pronominal] depending on whether the control is obligatory ornon-obligatory (arbitrary) (Williams 1980), see below. In obligatory control, PRO iscontrolled as it gets its meaning from another DP. In other words, PRO is bound by anantecedent that controls it and thus behaves like an anaphor as in (2) which is repeatedhere for convenience. As an anaphor, PRO must be bound in its binding category tosatisfy Principle A of binding theory.

(17) Johni tried [S’ [S PROi/*jto understand the problem]].

In non-obligatory control, PRO is a pronominal and is subject to Principle B (i.e. itshould be free in its governing category) of binding theory as in (18) where PRO takesarbitrary reference. There is a contradiction here as PRO cannot be free and bound in thesame domain. PRO Theorem requires that PRO be a pronominal and an anaphorsimultaneously. According to Government and Binding (GB), PRO Theorem proposesthat PRO does not have a governor, and thus does not have a governing category as italways appears in ungoverned positions (i.e. Spec of untensed TP which explains itsphonetically null status), so it is ungoverned.

(18) [PRO arb to help the poor], visit this website.

4. Control taxonomy

Two types of control have been proposed by Williams (1980): Obligatory and Non-obligatory control (OC and NOC). He argues that there are some properties pertinent toOC and that cases lacking these properties are classified NOC. In OC there must be alocal antecedent that precedes and c-commands PRO, PRO’s position cannot be filled byany other NP. PRO must be grammatically or semantically related to its antecedent. Hereare some examples to illustrate the difference between OC and NOC, adapted fromWilliams (1980) and Hornstein (1999).

28 A. Al-Kajela

Dow

nloa

ded

by [

104.

157.

129.

119]

at 1

7:13

29

July

201

6

(19) *It is expected PRO to dress himself. (There is no antecedent)(20) *John believes that it was expected to dress himself. (The antecedent John is not

local)(21) *John’s campaign expects PRO to shave himself. (The antecedent does not

c-command PRO)(22) *Johni told MaryjPROi+jto wash themselves. (OC cannot have split

antecedents)(23) John expects PRO to win and Bill does too. (OC permits sloppy interpretation

with ellipsis)

NOC has different properties; it is always following a complementizer:

(24) It was believed that PRO shaving was important. (NOC does not require anantecedent)

(25) Johni thinks that it is believed that PROi shaving himself is important.(Antecedent is not local)

(26) John thinks that PRO getting his work done is important and Bill does too.(NOC does not allow sloppy reading with ellipsis)

(27) Johni told Maryi that PROi+j washing themselves would be fun. (NOC permitssplit antecedents)

(28) John’si campaign expects that PROi raising funds at this time would beinappropriate. (In NOC, the antecedent does not c-command PRO)

(29) For John to stay would be our pleasure. (NOC permits lexical NP to appear inPRO position)

5. Theoretical issues

According to Hornstein (1999), the distribution of PRO is stipulated and the analysis ofnull Case in GB is not appealing: other null expressions such as wh-traces do not bearthis ‘tailored’ null Case. In addition, this kind of Case is only assigned to PRO by onekind of T. Lightfoot (1976) explains that NP-traces yield well-formed sentences withwant to contraction whereas wh-traces do not as in (30–5). This indicates that Chomskyand Lasnik’s (1993) argument for null-Case-marked PRO from non-finite T is violablebecause this empty category does not behave like other null categories. In fact, PROpatterns much more like an NP-trace, and thus should be treated as one (i.e. an NP-tracewithout null Case).

(30) Who do you want [t to go]?(31) *Who do you wanna go?(32) John is going [t to go].(33) John is gonna go.(34) I want [PRO to go].(35) I wanna go.

Furthermore, we can get the same sloppy interpretation with raising construction whichmakes perfect sense to claim that OC PRO is an NP-trace as well.

WORD 29

Dow

nloa

ded

by [

104.

157.

129.

119]

at 1

7:13

29

July

201

6

(36) John proved to be right and Bill did too (= the sentence is ambiguous: ‘John andBill each proved to be right’ or ‘John and Bill proved that John was right’).

Notice the explanation Chomsky and Lasnik provided for their claim in (37) that PRO isa null-Case marked category. They argue that PRO movement is a typical movementunder passive; it is a movement of the object of a passive verb where there is no Case.The non-finite specifier position can check the null Case assigned solely to PRO. In (38)PRO’s position as the Spec of non-finite T is appropriate, but the sentence is ill-formeddue to PRO’s movement out of Case-marked position.

(37) We never expected [PRO1 to be foundt1](38) *It is unfair [PRO1 to talk aboutt1]

As we mentioned earlier, GB tried very hard to establish the distribution of PRO; itappeared in the position of Spec of non-finite TP. Hornstein (1999) points out that PROcan occur in positions other than [Spec, TP] as in (39). When the object is overt, theanaphor will be spelled out. When it is covert, PRO will appear in object position as longas its antecedent has a ϴ-role.

(39) Johni washed (PROi/ himself).

6. Control as movement and the opponents

In GB theory, the postverbal NP of control verbs was base generated as a sister of thematrix verb as in (40). Lasnik and Saito (1991) propose that the NP his mother movesfrom the object position into Spec, Agro but still within the matrix clause, as shown in(41). This kind of analysis is analogous to that of raising with a difference lying in thekind of null category in question.

(40) John urged his mother [to meet his girlfriend].(41) John urged [AgrOP his mother i [VPti [PROi to meet his girlfriend]]].

Building on this state of affairs, Hornstein (1999) takes a crucial step to modify two of theprincipal assumptions of GB. His proposal is built on dispensing with (1) ϴ-criterion byallowing the arguments to bear more than one ϴ-role, (2) PRO Theorem. He claims thatthe Minimalist Program does not recognize the deep structure level of representation;therefore, it is quite appropriate to dispense with the restrictions that deep structurebrought with it. One of those restrictions is ϴ-positions. His theory builds on theobservation that OC or controlled PRO has anaphoric features and the NOC or arbitraryhas pronominal features.

Hornstein (1999), within a minimalist approach, tries to draw on this distinction toestablish a convergence correlation between OC and NP-traces to have a hybrid modelthat accounts for both raising and control in the same fashion. In his treatment of control,he considers ϴ-roles as morphological features on verbs (see Boskovic 1994 for similarproposal). By doing so, he allows NPs to move to a ϴ-position and receive a ϴ-role bychecking the thematic features of the verb. It is worth noting that Chomsky (1995) did notapprove the idea of having ϴ-roles as features. Hornstein’s proposal is not supporting theidea of grouping NPs as arguments, which have ϴ-roles by depending on their semantic

30 A. Al-Kajela

Dow

nloa

ded

by [

104.

157.

129.

119]

at 1

7:13

29

July

201

6

identity. “It is that ϴ-roles are features that DPs accrete through movement” (Hornstein2000: 134). He further states that “ϴ-roles are features and they can license movement”(135). On the contrary, verbs are grouped by their semantic identity, and thus it islegitimate for verbs to have ϴ-roles as features. The line of analysis he follows can bedescribed in the following configurations which are chiefly regulated by Rosenbaum’s(1967) Minimal Distance Principle (MDP), the first is set to account for subject controland the second for object control. The MDP is responsible for choosing the closestpotential antecedent to be the controller.

(a). NPi [V [PRO…]]

(b). NP [V NPi [PRO…]]

An argument will start off by checking the ϴ-role of the verb (merge) in the embeddedclause. Then, it moves to the [Spec, IP] of the lower clause, but this is not a Case-checking position. The argument moves again to check the ϴ-role of the matrix verb, itbears two ϴ-roles at this point. Accordingly, this characteristic feature teases apart controland raising construction whose subject would bear one ϴ-role from the embedded clause.The last move is to the [Spec, IP] of the matrix clause where the argument checks Casefeature and gets nominative Case.

Hornstein’s treatment of control as movement has been criticized by bothsyntacticians and semanticists such as Brody (1999), Landau (2003), Culicover andJackendoff (2001), Jackendoff and Culicover (2003). Landau (2003) argues thatHornstein fails to clearly account for passivized control verbs. If Hornstein’s movementtheory of control (MTC) is tenable, (42) should be derivable just like (43) as suggested byDavies and Dubinsky (2004). Hornstein (2000) addresses this issue raised by Brody andLandau by exploiting Chomsky’s (1998) notion of ‘phase’. He propounds that CPs arephases and for movement to be possible, the complement cannot be a CP when thatmovement occurs for control verbs such as hope and try. He further assumes that“complementizer incorporation can void the CP phase derivationally” (137), but thisoperation is blocked by passive in (42).

(42) * John was hoped to win the game.(43) John was expected to win the game.

He states that passive verbs as in (44) and (45) do not sanction complementizer deletion.However, this assumption as Davies and Dubinsky (2004) confirm, is untenable as thereare cases where the ‘that’ complementizer can be deleted after a passive verb as in (46).

(44) John fervently believes (that) there is a man here.(45) It’s fervently believed?? (that) there is a man here.(46) It was alleged (that) he took bribes from customers.

Culicover and Jackendoff (2001) have also criticized Hornstein’s MTC. Culicover andJackendoff’s main argument is that Hornstein’s theory fails to thoroughly cover the widerange of thematically distinct verbs. Their first observation is that Hornstein draws ananalogy, except for ϴ-roles, between PRO and anaphor with OC construction as in (47)and (48), but this does hold for other OC verbs such as try as in (49) and (50).

WORD 31

Dow

nloa

ded

by [

104.

157.

129.

119]

at 1

7:13

29

July

201

6

(47) John expects [PRO to win].(48) John expects himself to win.(49) John tried [PRO to win].(50) *John tried himself to win.

The other point raised by Culicover and Jackendoff (2001) is Hornstein’s manipulation ofMDP and the distinction between OC and NOC. Rosenbaum’s MDP stipulates that thecontroller is the closest NP to the infinitival clause. They employed Rosenbaum’s (1967)observation that with obligatory control predicates the transitive verbs constitute objectcontrol while the intransitive ones constitute subject control, as in (51) and (52). This alsoapplies to persuade-type verbs as in (53).

(51) Johni got to shave himselfi/*oneself.(52) Johni got Fredj to shave himselfj/*i / *oneself.(53) Johni persuaded Fredj to shave himselfj/*i/*oneself.

To further support his MDP, Rosenbaum used non-finite adjuncts to illustrate cases of OCwith subject being the controller as in (54–7).

(54) John flattered Mary without compromising himself/*herself.(55) John flattered Mary in order to vindicate himself/*herself.(56) John flattered Mary after (*Fred (’s)) injuring himself/*herself.(57) John flattered Mary while (*Fred (’s)) insulting himself/*herself.

However, MDP has its own shortcomings, as there are highly marked verbs such aspromise and vow which violate the MDP. These predicates are transitive and according toMDP, they should constitute object control construction as in (53). On the contrary, thesubject of the upper clause acts as the controller deviating from the norm as in (58–60).

(58) John promised to shave himself/*oneself.(59) Johni promised Fredj to shave himselfi/*j.(60) John promised Mary to leave.

In addition to Rosenbaum’s exception ‘promise’, Culicover and Jackendoff (2001) reportthat there is a large class of nominals which clearly violate the MDP by allowing subjectcontrol such as (promise, vow, offer, guarantee, pledge, oath, obligation), as shown in(61) and (62).

(61) John’s promise/vow/offer/guarantee/obligation/oath to Susan to take care ofhimself/*herself.

(62) John’s agreement/contract/arrangement with Susan to take care of himself/*herself.

Boeckx and Hornstein (2003) account to the anomalous behavior of the verb ‘promise’ byreferring to Chomsky’s (1969) observation that children acquire object control construc-tions earlier than subject control constructions.

32 A. Al-Kajela

Dow

nloa

ded

by [

104.

157.

129.

119]

at 1

7:13

29

July

201

6

Hornstein’s MTC would face another obstacle in accounting for verbs such as prayand plead and a large class of nominals such as invitation, instructions, order, reminder,and encouragement where the object of the preposition controls the subject of theinfinitival clause as in (63). Davies and Dubinsky (2004: 353) state that “the moved NPwould presumably fail to c-command its original position within the complement clause”.

(63) John prayed [to Athena] to take care of herself/*himself.

7. Arabic and super Equi

After this concise introduction about control in English, we turn now to one of theSemitic languages. Before getting into control in Standard Arabic, a short account of howthe grammar of this language works is in order. SA incorporates both SVO and VSO, butwe shall adopt SVO throughout this work for the sake of consistency. For more details onthis issue in Arabic, the reader is referred to El-Yasin (1985), and in English toMcCawley (1970). With the VSO, the verb must be in the singular, even when the subjectis in the plural or dual, whereas the SVO pattern requires full agreement with the subject.However, the verb always agrees in gender with the subject as shown in (64–73), forfurther discussion see Anshen and Schreiber (1968).

(64) ‘ʔal-muʕalɪm-u katab-a ʔa-ddarrs-a.’The teacher 3.s.m writes PERF.3rd.s.m the lessons.m.The male teacher wrote the lesson.(65) ‘ʔal-muʕalimu-na katab-u ʔa-ddarrs-a.’The teachers 3.pl.m write PERF.3rd.p.m the lessons.m.The teachers wrote the lesson.(66) ‘ʔa-lmuʕalima-tu katab-at ʔa-ddarrs-a.’The male teachers 3.pl.f write PERF.3rd.p.f the lessons.m.The female teacher wrote the lesson(67) *‘ʔal-muʕalimu-na katab-a ʔa-ddarrs-a.’The teachers 3.pl.m write PERF.3rd.s.m the lessons.m.The male teachers wrote the lesson.(68) ‘Katab-a ʔal-muʕalɪm-u ʔa-ddarrs-a.’Write PERF.3.s.m. the teacher 3rd.s.m the lesson s.m.The male teacher wrote the lesson.(69) *‘katab-u ʔal-muʕalimu-na ʔ-aldarrs-a.’Write PERF.3.pl.m the teachers 3rd.pl.m the lesson s.mThe male teachers wrote the lesson.(70) ‘Katab-a ʔal-muʕalimu-na ʔa-ddarrs-a.’Write PERF.3.s.m. the teachers 3rd.pl.m the lesson s.mThe male teachers wrote the lesson.(71) ‘Katab-at ʔal-muʕalima-tu ʔa-ddarrs-a.’Write PERF.3.s.f the teacher 3rd.pl.f. the lesson s.mThe female teachers wrote the lesson.(72) *‘katab-na ʔal-muʕalima-tu ʔa-ddarrs-a.’Write PERF. 3. pl.f. the teacher 3rd.pl.f. the lesson s.mThe female teachers wrote the lesson.

WORD 33

Dow

nloa

ded

by [

104.

157.

129.

119]

at 1

7:13

29

July

201

6

(73) ‘Katab-at ʔal-muʕalima-tan ʔa-ddarrs-a.’Write PERF. 3. s.f. the teacher 3rd.du.f. the lesson s.mThe two female teachers wrote the lesson.

It is a subject pro-drop language where the pronoun has a non-arbitrary reference likeEnglish and it refers to a definite first, second, or third person. Pro is a silent element,which can alternate with a lexical DP as in (74, 76, 78, and 80). However, pronouns areusually used to indicate either emphasis or contrast as shown in (75, 77, 79, and 81).Nouns are inflected for gender, number, and person. Verbs are inflected for two aspects:perfective and imperfective, which do not express the time of an event in the same way asthe primary tenses in Indo-European languages, and they do not correspond to themeaning of these terms in English.

(74) ‘Pro ʔa-qraʔ-u l-kɪtaab-a.’1. s.m/f read IMP the bookI read the book.(75) ‘ʔana ʔa-qraʔ-u l-kɪtaab-a.’I 1.s.m/f read IMP the bookI read the book.(76) ‘Pro ta-qraʔ-u l-kɪtaab-a.’2. s.m read IMP the bookYou read the book.(77) ‘ʔanta t-aqraʔ-u l-kɪtaab-a.’You 2.s.m read IMP the bookYou read the book.(78) ‘Pro ya-qraʔ-u l-kɪtaab-a.’3. s.m read IMP the bookHe reads the book.(79) ‘Huwa ya-qraʔ-u l-kɪtaab-a.’He 3.s.m read IMP the bookHe reads the book.(80) ‘Pro ta-qraʔ-u l-kɪtaab-a.’3. s.m read IMP the bookShe reads the book.(81) ‘Hiya ta-qraʔ-u l-kɪtaab-a.’She 3.s.f read IMP the bookShe reads the book.

Here, we try to investigate whether SA recognizes OC and NOC in its grammar the sameway English does, and to achieve this purpose we will employ some of William’s (1980)diagnostics. In (82) which is supposed to be the counterpart of English ellipsis (26), butunexpectedly we get the sloppy interpretation which is the by-product of NOC. The verbin the embedded clause is in the subjunctive mood, inflected for third person singular, andcoindexed with the subject of the upper clause ‘ʕaliyy-un’ and the subject of theembedded clause ‘ʔaħmad-un’. The sloppy interpretation is still present even when theverb is coindexed with third person feminine in the matrix clause as in (83).

34 A. Al-Kajela

Dow

nloa

ded

by [

104.

157.

129.

119]

at 1

7:13

29

July

201

6

(82) ‘Tawqaʕ-a ʕaliyy-un ʔan ja-fuz-a w- ʔaħmad-u ʔejḍan.’Expect.PERF.3s.m. Ali Comp IMP. win3.s.m. and Ahmad too.Ali expected to win and Ahmad did too.(83) ‘Tawqaʕ-at Suaad-u ʔan ta-fuz-a w- ʔaħmad-u ʔejḍan.’Expect.PERF.3s.f. Suaad Comp IMP.win3.s.f. and Ahmad too.Suaad expected to win and Ahmad did too.

Arabic language permits split antecedents, which is another sign in NOC direction.However, Arabic morphology behaves in a peculiar way; the embedded predicate iscoindexed with the matrix subject gender-wise as it is inflected for dual masculine genderas in (84). The predicate is coreferential with both subject and object of the matrix clause.The predicate of the embedded clause will keep its masculine inflection whether thesubject of the matrix clause is masculine or feminine. The coreferentiality stipulates thatboth the subject and object of the matrix clause control the subject of the embeddedclause. It is worth noting that the feminine inflection would be used only when botharguments are females, which has to do more with pragmatics than syntax andmorphology.

(84) ‘ʕaliyy-uni ʔaqnaʕ-a suʕaad-aj ʔan ja-Ɣtasil-a pro *i/*j/i+j.’Ali persuade.PERF.3.s.m Suaad Comp IMP.wash.3.m.du.Ali persuaded Suaad to wash themselves.(85) ‘suʕaad-ui ʔaqnaʕ -at ʕaliyy-anj ʔan ja-Ɣtasil-a pro *i/*j/i+j.’

Suaad persuade.PERF.3.s.f Ali Comp IMP. wash. 3.m.duSuaad persuaded Ali to wash themselves.

Further evidence that Arabic grammar does not license OC constructions comes fromadjuncts. Hornstein (1999) noticed, while trying to eliminate PRO, that with PRO-headedadjuncts the controller of PRO violates MDP as in (86). In addition, the MDP incorrectlypredicted that control into adjunct clauses have only subject control and thus excludedobject control. In fact, this problem was due to MDP, PRO, and the assumption thatobjects cannot bind into adjunct (Chomsky 1995: 272). Hornstein pointed out that“objects can bind into adjuncts” (Hornstein 1999: 76).

(86) Johni saw Maryj without PROi/*j leaving the room.(87) Mary greeted every boyi without hisi knowing it.

Arabic grants more support to PRO elimination as it allows control into adjuncts clauseswhich have both subjects and objects. It is worth noting that a lexical DP cannot appearafter complementizer ʔan and hence pro is also banned in this position as in (88 and 89).

(88) ‘ʕaliyy-uni ʔanqaða suʕaad-anj min-duuni ʔan ta-ʕrif-a pro*i/j.’Ali save.PERF.3.s.f Suaad without Comp IMP. know 3.s.f.Ali saved Suaad without (her) knowing it.(89) ‘ʕaliyy-uni ʔanqaða suʕaad-anj min-duuni ʔan ja-ʕrif-a proi/*j.’Ali save.PERF.3.s.f Suaad without Comp IMP. know 3.s.m.Ali saved Suaad without (him) knowing it.

WORD 35

Dow

nloa

ded

by [

104.

157.

129.

119]

at 1

7:13

29

July

201

6

These are supposed to be canonical object control verbs. The ambiguity in (90) emergesfrom the fact that the embedded null subject can be coindexed with the matrix subject orobject. In sentence (91), the object of the embedded clause is an anaphor, but the sentenceis still ambiguous as it can be interpreted to be coindexed with either the subject or theobject of the matrix clause. This ambiguity can be partly attributed to the thematicfeatures of the matrix verb, because changing the matrix will resolve some of thisambiguity by having either the object or the subject of matrix clause as the controller ofthe embedded subject as in (92) and (93). The ambiguity is still unresolved in (94) wherethe embedded null subject can still be coindexed with the matrix subject and object.

It is also noticed that antecedent choice for the embedded subject violatesRosenbaum’s (1967) MDP which proposes that the controller in obligatory controlconstructions must c-command the null subject of the clausal complement of the controlpredicate.

(90) ‘ʕaliyy-uni ʔɪttafaq-a maʕ ʔaħmad-aj ʔan ju-ṣawir-a-hui/j.’Ali agree.PERF 3.s.m with Ahmad that IMP.3.m-photograph-him.Ali agreeed with Ahmad to photograph him (Ali or Ahmad).(91) ‘ʕaliyy-uni ʔɪttafaq-a maʕ ʔaħmad-aj ʔan ju-ṣawir-a nafsa-hui/j.’Ali agree.PERF 3.s.m with Ahmad that IMP.3.m-photograph self-him.Ali agreed with Ahmad to photograph himself (Ali or Ahmad).(92) ‘ʕaliyy-uni ʔamar-a ʔaħmad-aj ʔan ju-ṣawir-a nafsa-hu*i/j.’Ali order.PERF.3.s.m Ahmad Comp IMP.3.m.photograph self-him.Ali ordered Ahmad to photograph himself (Ahmad).(93) ‘ʕaliyy-uni ʔamar-a ʔaħmad-aj ʔan ju-ṣawir-a-hui/*j.’Ali order.PERF.3.s.m Ahmad Comp IMP.3.m.photograph-him.Ali ordered Ahmad to photograph him (Ali).(94) ‘ʕaliyy-uni ʔamar-a ʔaħmad-aj ʔan ju-ṣawir-a axa-hui/j.’Ali order.PERF.3.s.m Ahmad Comp IMP.3.m.photograph brother-his.Ali ordered Ahmad to photograph his brother (Ali’s or Ahmad’s brother).

Nominals in English have proved a case of violation against MDP by allowing subjectcontrol where the controller is not c-commanding the controlee. In Arabic, this is not thecase due to the rich inflectional morphology. The object is blocked because the embeddedverb is inflected for masculine and thus solely coindexed with the subject of the upperclause, as shown in (95). However, this can be solved if the embedded verb changes itsmasculine morphological status to be coindexed with the object only (96).

(95) ‘ʔittifaq-u ʕaliyy-in maʕ suʕaad-in ʔan ja-htamm-a bi-nafsi-hi/ *bi-nafsi-ha.’Agreement-POSS Ali with Suaad Comp IMP.3.m-take care in-self-him in-self-herAli’s agreement with Suaad to take care of himself.(96) ‘ʔittifaq-u ʕaliyy-in maʕ suʕaad-in ʔan ta-htamm-a *bi-nafsi-hi/ bi-nafsi-ha.’Agreement-POSS Ali with Suaad Comp IMP.3.f-take care in-self-him in-self-herAli’s agreement with Suaad to take care of herself.

It is worth noting that SA exhibits a sole case of obligatory control with verbal nouns asin (97). But even this case vanishes when the complementizer ʔan is introduced to forman embedded clause as in (98).

36 A. Al-Kajela

Dow

nloa

ded

by [

104.

157.

129.

119]

at 1

7:13

29

July

201

6

(97) ‘Proħawal-a l-qitaal-a.’Try PERF.3.s.m the fightingHe tried to fight.(98) ‘Proħawal-a ʔan ju-qatil-a.’TryPERF.3.s.m Comp IMP.3.m-fightHe tried to fight.

8. Concluding remarks

After applying most of William’s (1980) diagnostics, the examples have argued for NOCin SA. This can be attributed to (1) SA does not have non-finite clauses like those ofEnglish. The closest embedded clauses to English are the subjunctive ones and those areusually preceded by a complementizer similar to ‘that’. (2) SA has an intricatemorphological system which inflects verbs, nouns, and adjectives for person, number,and gender. (3) Instead of PRO, SA has pro which occupies the position of the nullsubject in the embedded clause. Hornstein (1999: 92) indicates that OC and NOC are in“complementary distribution”. The bottom line is that in order for control to survive,especially the proposed OC which involves A-movement, one should take into hisaccount a broader framework; a hybrid framework which can saturate syntax, semantics,and sometimes pragmatics.

AcknowledgmentsI would like to thank Dr. El Khachab from York University for his comments on some Arabicexamples. I am also grateful to the two anonymous reviewers for their technical and editingcomments.

Disclosure statementNo potential conflict of interest was reported by the author.

REFERENCESAnshen, Frank & Peter A. Schreiber. 1968. A focus transformation of modern standard Arabic.

Language 44. 792–7.

Boeckx, Cedric & Norbert Hornstein. 2003. Reply to “control is not movement”. Linguistic Inquiry34. 269–80.

Boskovic, Zeljko. 1994. D-structure, theta criterion, and movement into theta positions. LinguisticAnalysis 24. 247–86.

Brody, Michael. 1999. Relating syntactic elements: Remarks on Hornstein’s “Movement andchains”. Syntax 2. 210–26.

Chomsky, Carol. 1969. The acquisition of syntax in children from 5 to 10. Cambridge, MA: MITPress.

Chomsky, Noam. 1965. Aspects of the theory of syntax. Cambridge, MA: MIT Press.

Chomsky, Noam. 1981. Lectures on government and binding. Dordrecht: Foris.

WORD 37

Dow

nloa

ded

by [

104.

157.

129.

119]

at 1

7:13

29

July

201

6

Chomsky, Noam. 1995. The minimalist program. Cambridge, MA: MIT Press.

Chomsky, Noam. 1998. Minimalist inquiries: The framework. MIT Occasional Papers inLinguistics 15. Cambridge, MA: MIT Working Papers in Linguistics.

Chomsky, Noam & Howard Lasnik. 1993. The theory of principles and parameters. In JoachimJacobs, Arnim von Stechow, Wolfgang Sternefeld & Theo Vennemann (eds.), Syntax: Aninternational handbook of contemporary research, 506–69. Berlin: Walter de Gruyter.

Culicover, Peter & Ray Jackendoff. 2001. Control is not movement. Linguistic Inquiry 32.493–512.

Culicover, Peter & Wendy Willinks. 1986. Control, PRO, and the projection principle. Language62. 120–53.

Davies, William & Stanley Dubinsky. 2004. The grammar of raising and control: A course insyntactic argumentation. Oxford: Blackwell Publishing.

El-Yasin, Mohammed Khalid. 1985. Basic word order in classical Arabic and Jordanian Arabic.Lingua 65. 107–22.

Hornstein, Norbert. 1997. Control in GB and Minimalism. Glot International 8.2. 3–6.

Hornstein, Norbert. 1999. Movement and Control. Linguistic Inquiry 30. 69–96.

Hornstein, Norbert. 2000. On A-chains: A reply to Brody. Syntax 31. 29–43.

Jackendoff, Ray & Peter Culicover. 2003. The semantic basis of control in English. Language 79.517–56.

Landau, Idan. 2003. Movement out of control. Linguistic Inquiry 34. 471–98.

Lasnik, Howard & Mamoru Saito. 1991. On the subject of infinitives. Proceedings of the ChicagoLinguistic Society (CLS) 27. 324–43.

Lightfoot, David. 1976. Trace theory and twice-moved NPs. Linguistic Inquiry 7. 559–82.

McCawley, J. D. 1970. English as a VSO language. Language 46. 286–99.

Rosenbaum, Peter S. 1967. The grammar of English predicate complement constructions.Cambridge, MA: MIT Press.

Williams, Edwin. 1980. Predication. Linguistic Inquiry 11. 203–38.

38 A. Al-Kajela

Dow

nloa

ded

by [

104.

157.

129.

119]

at 1

7:13

29

July

201

6