Agreement asymmetries in Arabic varieties dissolved: A feature inheritance-based account

Abstract

The paper revisits the issue of the structural representation of subject-verb agreement in Arabic varieties with a particular reference to Standard Arabic and seven Arabic dialects, i.e., Saudi Arabic, Yemeni Arabic, Iraqi Arabic, Jordanian Arabic, Tunisian Arabic, Moroccan Arabic, and Lebanese Arabic. The paper recasts agreement asymmetries proposed in the literature of Arabic syntax: partial agreement in VSO but full agreement in SVO. Empirical evidence shows that such proposed asymmetries should be dissolved as all language varieties display agreement between the verb and the subject in all features except the number feature in the VS order of standard variety. Even in the VS order of this variety, the verb fully agrees with the subject when the latter is a pronoun. Based on empirical observations, the puzzling phenomenon of agreement can receive a straightforwardly unified account based on the Feature Inheritance model of syntax (FI). Irrespective of the possible order used in Arabic, full agreement morphology is always predicted as a result of an Agree relation, via FI, between the probe T and the subject generated at Spec; vP. However, the number feature alternation in the VS order of Standard Arabic is attributed to the PF interface rather than the narrow syntax, i.e., the possibility to spell out the number feature periphrastically rather than affixially.

Keywords:

• agreement
• asymmetry
• merger
• word order
• feature inheritance
• derivation

PUBLIC INTEREST STATEMENT

Subject-verb Agreement is one of the interesting phenomena across languages. The verb may match with the subject in some or all nominal features (person, number, and gender). In Standard Arabic, full agreement is morphologically shown in SVO but partial agreement in VSO. The main observation of the paper is that the proposed agreement–word order asymmetries in Standard Arabic are boiled down in modern dialects of Arabic, i.e., the verb fully agrees with the subject irrespective of the word order (VSO or SVO). Besides, the verb in Standard Arabic fully agrees with the subject in the VSO order when the subject is a pronoun. The agreement facts call for revisiting the previous analyses rested on such agreement-word order asymmetries. The study develops a unified morphosyntactic account for agreement facts in standard and dialectal varieties of Arabic so that the learning difficulties faced by non-native speakers of Arabic are adequately resolved.

1. Introduction

Agreement, for Baker (2008), is the grammatical operation by which the form of one item forces a second item in the sentence to appear in a particular form’. It refers to “a syntactic process whereby one constituent must have the same value for certain grammatical features (such as person, number, and gender) as another constituent that it bears a particular grammatical relation to” (Frankin et al., 2000: 684). That is, it is a formal relationship between two constituents in a certain hierarchical structure where the form of one constituent requires from the other to match with. It occurs between two elements in a sentence, i.e., between determiner and argument, between adjective and argument, and verb and argument. However, it is the latter that the paper exclusively focusses on. In subject-verb agreement, the verb may match with the subject in some or all nominal features (person, number, gender), oblivious to the presence of any other noun phrase in the sentence.

In Standard Arabic, the subject may precede or follow the verb, yielding two possible word orders: VSO and SVO. Several proposals in the literature of Arabic syntax (Mohammad, 1990; Fassi-Fehri, 1993; Aoun et al., 1994; Ouhalla, 1994; Soltan, 2007; Al-Horais, 2009; among many others) sought to account for agreement patterns in Standard Arabic by resting on their assumption that there is an asymmetric relation between agreement pattern and word order alternation: partial agreement in VSO but full agreement in SVO. Examples (1) and (2) from Standard Arabic illustrate these language facts.

Table

1)	a.	qaraʔa	l-ʔawlaad-u	l-qiṣat-a	VSO = Partial Agreement
		read.3^rd.SG.M	the-boys-Nom	the-story-Acc
	b.	*qaraʔ-u	l-ʔawlaad-u	l-qiṣat-a
		read.3^rd.PL.M	the-boys-Nom	the-story-Acc
		“The boys read the story.”

Table

2)	a.	ʔal-ʔawlaad-u	qaraʔ-u	l-qiṣat-a	SVO = Full Agreement
		the-boys-Nom	read.3^rd.PL.M	the-story-Acc
	b.	*ʔal-ʔawlaad-u	qaraʔa	l-qiṣat-a
		the-boys-Nom	read.3^rd.SG. M	the-story-Acc
		“The boys read the story.”

In (1a) given for the VS order, there is no full agreement between the postverbal subject l-ʔawlaad-u “the-boys-Nom” and the verb qaraʔ-a “read.3^rd.Singular.Masculine”. The verb does not agree with the subject in number. By contrast, there is full agreement (person, gender, and number) between the preverbal subject ʔal-ʔawlaad-u “the-boys-Nom” and the verb qaraʔ-u “read.3^rd.Plural.Masculine” in (2a) given for the SV order. Sentences (1b) and (2b) are grammatically starred due to the fact that neither full agreement in VS nor partial agreement in SV is permissible in Standard Arabic.

Now, let us examine whether the agreement asymmetry contingent on word order applies in other varieties of Arabic, namely (Najdi) Saudi Arabic as in (3), (Sana’ani) Yemeni Arabic as in (4), Jordanian Arabic as in (5), Iraqi Arabic as in (6), Tunisian Arabic as in (7), Moroccan Arabic as in (8), and Lebanese Arabic as in (9).

Table

3)	a.	ʃa:f-u:	ʔil-ʕia:l	ʔil-film
		watched.3rd.PL.M	the-boys	the-film
	b.	ʔil-ʕia:l	ʃa:f-u:	ʔil-film
		the-boys	watched.3rd.PL.M	the-film
		‘The boys watched the movie.’
	Najdi Saudi Arabic (Alwahibee:2020:10)

Table

4)	a.	ragad-ū	ʔil-ʤaħāl
		slept.3rd.PL.M.	the-children
	b.	ʔil-ʤaħāl	ragad-ū.
		the-children	slept.3rd.PL.M
		‘The children slept.’
	Data collected from native speakers of (Sana’ani) Yemeni Arabic

Table

5)	a.	wisˤl-u	l-iwlaad
		arrived.3rd.PL.M	the-boys
	b.	l-iwlaad	wisˤl-u
		the-boys	arrived.3rd.PL.M
		‘The boys arrived.’
	Jordanian Arabic data adapted from Jarrah(2017:6)

Table

6)	a.	Kitbaw	al-wilad	al-qissa
		wrote.3rd.PL.M	the-boys	the-story
	b.	al-wilad	Kitbaw	al-qissa
		the-boys	wrote.3rd.PL.M	the-story
		‘The boys wrote the story.’
	Iraqi Arabic data adapted from Hazem (2017:367)

Table

7)	a.	kitb-uu	T-Tulaab	d-dars
		wrote.3rd.PL.M	the-students	the-lesson
	b.	T-Tulaab	kitb-uu	d-dars
		the-students	wrte.3rd.PL.M	the-lesson
		‘The students wrote the lesson.’
	Tunisian Arabic data adapted from Jouini (2018:234)

Table

8)	a.	nəʕs-u	l-əwlad
		slept.3rd.PL.M	the-children
	b.	l-əwlad	nəʕs-u
		the-children	slept.3rd.PL.M
		‘The children slept.’
	Moroccan Arabic data adapted from Aoun et al (2010:84)

Table

9)	a.	neem-o	l-wleed
		slept.3rd.PL.M	the-children
	b.	l-wleed	neem-o
		the-children	slept.3rd.PL.M
		‘The children slept.’
	Lebanese Arabic data adapted from Aoun et al (2010:84)

Data from Arabic dialects show that the verb should fully agree with the subject in any possible word order. In Saudi Arabic, as shown in (3), the verb ʃa:f-u: “watched.3^rd.Plural.Masculine” needs to fully agree with the subject ʔil-ʕia:l “the-boys” regardless of its position: postverbally as in (3a) or preverbally as in (3b). Similarly, Yemeni Arabic, as illustrated in (4), requires the verb ragad-ū “slept.3^rd.Plural.Masculine” to have full agreement features with the subject ʔil-ʤaħāl “the-children.Masculine” irrespective of word order used. The case is similar in Jordanian Arabic in which the verb wisˤl-u “arrived.3^rd.Plural.Masculine”, as shown in (5), should fully agree with the subject l-iwlaad “the-boys” whether the subject precedes or follows the verb. Likewise, the verb kitbaw “wrote.3^rd.Plural.Masculine” in Iraqi Arabic must show full agreement morphology with the subject al-wilad “the-boys” as in (6) and this agreement pattern applies in any word order. In Tunisian Arabic, the verb kitb-uu “wrote.3^rd.Plural.Masculine” as illustrated in (7) needs to agree with the subject T-Tulaab “the-students-masculine” in all features, i.e., number, person, and gender, irrespective of a position: postverbally as in (7a) or preverbally as in (7b). Data from Moroccan Arabic and Lebanese Arabic also warrant the same conclusion. In Moroccan Arabic, the verb nəʕs-u “slept.3^rd.Plural.Masculine” must have full agreement with the subject l-əwlad “the-children-Masculine” regardless of the order. Similarly, the verb neem-o “slept.3^rd.Plural.Masculine” in Lebanese Arabic, as given in (9), fully agrees with the subject l-əwlad “the-children.Masculine” in any word order used. To sum up, full agreement between the verb and the subject is always manifested in all Arabic varieties, except in the VS order of Standard Arabic, and these facts are clearly shown in table (10).

Table

10)	Arabic Variety	Word order	Subject-Verb Agreement Features			Pattern
			Gender	Person	Number
	Standard Arabic	VSO	√	√	X	Partial Agreement
		SVO	√	√	√	Full Agreement
	Saudi Arabic	VSO	√	√	√	Full Agreement
		SVO	√	√	√	Full Agreement
	Yemeni Arabic	VSO	√	√	√	Full Agreement
		SVO	√	√	√	Full Agreement
	Jordanian Arabic	VSO	√	√	√	Full Agreement
		SVO	√	√	√	Full Agreement
	Iraqi Arabic	VSO	√	√	√	Full Agreement
		SVO	√	√	√	Full Agreement
	Tunisian Arabic	VSO	√	√	√	Full Agreement
		SVO	√	√	√	Full Agreement
	Moroccan Arabic	VSO	√	√	√	Full Agreement
		SVO	√	√	√	Full Agreement
	Lebanese Arabic	VSO	√	√	√	Full Agreement
		SVO	√	√	√	Full Agreement

Bearing in mind these facts, the crucial question here is how the previous analyses, which have been rested on the assumption of the working agreement-word order asymmetries of the standard variety, can capture agreement facts in other Arabic varieties. Hence, the main challenge of the paper is to provide a unified explanation of the puzzling agreement phenomenon in both standard and dialectical varieties of Arabic.

2. Literature review

Several analyses have been proposed in the literature of Arabic syntax (Mohammad, 1990, 2000; Fassi-Fehri, 1993; Aoun et al., 1994; Olarrea (1995); (Benmamoun, 2000; Soltan, 2007; Fakih, 2016; Alahdal, 2021) among others in attempt to provide possible accounts for subject-verb agreement based on word order asymmetries: partial agreement in VSO but full agreement in SVO. These accounts include the Null Expletive Analysis (Mohammad, 1990, 2000; Ouhalla, 1994; Soltan, 2007; Al-Horais, 2009), PF-Merger Analysis (Benmamoun, 2000), Agreement Loss Analysis (Aoun et al., 1994), Base-generated Analysis (Soltan, 2007), and Feature Inheritance-based Analysis (Fakih, 2016). I will briefly review these previous analyses to decide whether they can provide a plausible answer to the question of how agreement facts in standard Arabic and Arabic dialects are uniformly derived.

2.1. Null expletive analysis

Mohammad (1990, 2000), building on Pollock’s (1989) split Infl-system, proposes that both types of agreement (partial agreement in VSO and full agreement in VSO) are uniformly obtained under Spec-head configuration where the verb occupying the head position Infl, T in minimalist terms, enters into a checking relation with the element occupying the Spec; TP. Full agreement is manifested when Spec; TP is occupied by a lexical subject but partial agreement when it is occupied by a null expletive pro(nominal). This analysis cannot capture agreement facts of Arabic dialects simply because full agreement is obtained whether Spec; TP is occupied by a lexical subject or null expletive. In addition, null expletive hypothesis is weakened by Bahluol & Harbert’s (2002) counterargument that the verb in Standard Arabic needs also to fully (not partially) agree with a pronominal subject in the VS order as illustrated in (11).

Table

11)	a.	xaraj-uu	hum	wa	Sami
		left.3rd.PL.M	they	and	Sami
	b.	*xaraj-a	hum	wa	Sami
		left.3rd.SG.M	they	and	Sami
		“They and Sami left.”
	Standard Arabic; Bahluol & Harbert, 2002:64)

Bahloul and Harbert’s (2002) counterexamples provide an important insight that it further reduces the VS order domain of partial agreement in Standard Arabic into the context whereby the subject is not a pronoun. That is to say, all Arabic varieties manifest the full agreement morphology between the verb and the subject when placed in any word order, except in the VS order of the Standard Arabic with non-pronoun subjects.

2.2. PF merger analysis

Benmamoun (2000) proposes a morphological analysis for agreement asymmetry based on word order under Spec-head configuration. Agreement asymmetry in Standard Arabic, he argues, is exclusively attributed to the Spell-out of the number feature at PF: affixally (verb plus affix) in SVO; or periphrastically (verb plus postverbal subject) in VSO. The verb carries same number features of the subject, i.e., the verb always gets a plural agreement feature when the subject is plural. By virtue of the absence of PF merger between the subject and the verb in SVO, agreement (number) features appear affixially on the verb. By contrast, agreement (number) features do not appear on the verb in VSO due to the presence of a PF merger between the subject and the verb, the operation that makes the realization of number affix on the verb redundant. Redundancy is resulted because the subject is already specified for the same feature. Interestingly, the difference between Standard Arabic and Moroccan Arabic in terms of agreement, he claims, is due to a possible parameter that a variety may select: Moroccan Arabic has one option (the number agreement affixation on the verb), whereas Standard Arabic has two options (affixation and merger between the verb and the subject). However, this analysis cannot be structurally embraced. It rests on Spec-head configuration, a relation which was abandoned in recent minimalist theories and successfully replaced by an Agree relation in a Probe-Goal domain.

2.3. Agreement loss analysis

Aoun et al. (1994) claim that Standard Arabic has initially full agreement pattern in both SVO and VSO orders, and it is achieved under a Spec-head configuration. The number feature agreement is lost when the verb moves past the subject to derive the VS order. The analysis cannot account for similar facts of Arabic dialects whereby T-to-Focus movement does not lead to the loss of the number feature agreement on the verb. Moreover, the Spec-head configuration, on which the analysis was built, has been eliminated and effectively replaced by the long-distance Agree in the current minimalist framework.

2.4. Null pro analysis

Soltan (2007) proposes, in tandem with the proposals made for other null subject languages (Alexiadou & Anagnostopoulou, 1998; Solà, 1992; among others) that a preverbal DP in Standard Arabic is not a subject but a topic with A-bar properties. There are two types of subject in Standard Arabic: lexical DP in VSO and a null pro in SVO. The lexical subject that appears in SVO is not a subject but a topic. He claims that full agreement in SVO is obtained because T establishes an Agree relation with a null pro (not a lexical subject) to be licensed in line with Rizzi’s (1986) pro Identification Principle.¹ Conversely, partial agreement in VSO is manifested because T enters an Agree relation with the lexical DP. This account, though interesting in following an Agree approach, cannot account for the agreement data of Arabic dialects simply because Agree relation between T and the lexical subject in VSO yields full agreement and not partial agreement as he expects. See, among others, Al-Horais (2009), Fakih (2016), and Alahdal and Van De Weijer (2021) for theoretical and conceptual issues the account may encounter.

2.5. Feature inheritance-CP split analysis

Fakih (2016), combining essentially Chomsky (2008) Feature Inheritance with Rizzi’s (1997) Split-CP System, argues that the subject in VSO remains in situ (at Spec; vP), but it moves in SVO to a higher specifier position, i.e., Spec; TopP via TP, to accommodate topicalization. Partial agreement in VSO is obtained by establishing an Agree relation between the would-be probe T and the subject at Spec; vP. Full agreement in SVO is obtained, however, by establishing two simultaneous Agree relations via feature inheritance: One between the probe T and the subject at Spec; vP for valuation of φ-features; and the other between the probe Top and the subject at Spec; vP for the edge feature valuation. The analysis, I believe, is considered to be a first attempt to explain how agreement asymmetry in Standard Arabic is syntactically derived under one of the most recent minimalist approaches. However, it encounters some theoretical and conceptual issues. First, the analysis does not clarify how partial agreement is predicted as nothing in the proposed analysis may block full agreement to appear in VSO, contrary to the facts of Standard Arabic. This seems to be a failure to respond to the main question of why it is partial agreement in VSO as both of the agreement patterns are dealt alike under this analysis. Second, splitting of phasal probing features (φ-features inherited by T and Edge feature by ToP) in SVO is left unjustified. Third, the account does not take the SV as a possible word order in the language but as being contingent with topicalization, contrary to the language facts. Finally, I see no point of advantage for the assumption that a topicalized element is hosted by Spec; ToP rather than Spec; CP as assumed by Chomsky (2008). Moreover, postulating a Topic Phrase between the phase head C and its canonical nonphase head T should cast doubt on whether the main essence of feature inheritance, namely C-T relation, retains after positing the intervening projection (Topic Phrase) in between.

2.6. Lexical-morphosyntactic analysis

Alwahibee (2020) postulates that the agreement asymmetry in standard and dialectal varieties of Arabic is derived by interactions of Agree, constraints on the head T’s φ-probe, and postsyntactic requirements. That is, the probe T in Standard Arabic incorporates the feature geometry [uPart, uInd, uClass, EPP] in SVO but [uPart, vSG, uClass] in VSO. He claims that the number feature in VSO is lexically specified on the verb as valued singular [vSG] resulting in partial agreement on the verb (technically T) in the course of syntactic derivation. The analysis is ridden with some critical questions, which are left unanswered such as: What determines the selection of the correct alternative in the lexicon: valued singular feature in VSO and unvalued number in SVO? Isn’t based on a Look-Ahead step? Can an incomplete φ-feature bundle work under Agree effectively? The analysis brings more theoretical apparatus into the syntactic computation and this turns out to be disfavored in the minimalist framework which seeks elegance, economy, and beauty to represent the perfect design of the Language. Moreover, it breaks the language/dialect dichotomy internally as it claims that the selection of parametric feature geometry options selected for the probe T is based on specific properties of languages/dialects.

3. A feature inheritance-based alternative

I propose that the puzzling agreement phenomenon in all Arabic varieties can be uniformly captured under Chomsky’s (2008) Feature Inheritance model of syntax (henceforth FI), the mechanism conceptually advocated by Richards (2007). Basically, the Feature Inheritance approach reinterprets the relation between functional heads C and T as the head T does not have its own Agree φ-features and thus cannot serve as a probe. The head T becomes a probe only when it inherits probing features (φ-features) from the phase head C. That is, the head T cannot initiate any Agree operations directly or independently of C (the locus of φ-features). Traditional subject-verb agreement and EPP (Extended Projection Principle) effects associated with T (A-movement of the formal subject to Spec; T and expletives, etc.) can arise via FI mechanism, whereby uninterpretable probing features (with/without EPP) are passed down from the phase head C to its complement head T. According to Chomsky (2008), the movement to a specifier position of a given phrase is triggered by the Edge Feature (EF) on the phase head, which corresponds to the EPP feature. The features inherited to T can trigger subject A-movement to Spec TP, rather than Spec CP, since C’s Agree/φ-features have been passed onto T (φ-agreement with the subject can also spell-out via T, e.g., “She runs.”). Feature Inheritance mechanism can operate and account for the parametric variation of word order in a principled manner. Richards (2007) argues that feature inheritance is motivated by Full Interpretation, more specifically, the need to eliminate uninterpretable features (uFs) before they reach LF. He claims that the elimination of uFs is possible only when valuation and transfer (spell-out to PF) of uFs apply simultaneously. The projections of non-phase heads can only be transferred at the phase level; non-phase heads cannot have any uFs. Feature inheritance, he argues, is needed: uFs must spread from phase heads to non-phase heads (from C to T, v* to V, etc.).

In line with Chomsky’s (2008) Feature Inheritance, I propose that the nonphase head T is not a probe by itself but rather inherits its probing features, namely φ-features (and the Edge feature), from the phase head C. Via FI, the probe T enters in an Agree relation with the Goal, i.e., the subject at Spec; vP,² whereby the uninterpretable features of both matching elements get valued in the course of derivation, yielding full agreement morphology on the verb irrespective of the word order used. Since full agreement morphology can be manifested in any possible order (SV or VS) in Arabic, the previous analyses, which are based on agreement-word order asymmetries, cannot work. Moreover, deriving a possible word order depends on the presence or absence of the Edge Feature, i.e., EPP, on the proxy head T and has nothing to do with (dis)appearing agreement morphology. In Arabic SV derivation, the probe T, via FI, inherits the Edge Feature together with φ-features from the phase head C. The EF on the probe T is thus satisfied by raising the elements agreeing in φ-features with T to its edge as schematically shown in (12a). By contrast, in Arabic VS derivation, the Edge feature is not present on the Probe T as illustrated in (12b).

short-legend12)

In proposed structures, (12a) for the SV order and (12b) for the VS order, the nonphase head T becomes a probe by virtue of inheriting uninterpretable features (uφ-features) from the phase head C. The mere difference between the two derivations lies in the assumption that the Edge Feature (along with uφ-features) is shipped down to the head T in the former but not in the latter. In both derivations, the probe T, via FI, enters in an Agree relation with a matching goal, i.e., the subject at Spec; vP, whereby uninterpretable features on both sides get valued, yielding full agreement paradigm in the narrow syntax. In the SV derivation, unlike the VS derivation, the subject at Spec; vP needs to raise to Spec; TP to satisfy the EF requirement of the probe T. It is not the case in the VS derivation where the probe T does not trigger the subject movement due to the lack of the Feature, thus leaving the subject in situ.

3.1. Full agreement in VS derivation: C →T_φ-Features

In the derivation of VS sentences, the phase head C passes down its probing features, namely unvalued φ-features, to its complement head T. Being a probe by virtue of FI, T enters into an Agree relation with a matching goal, i.e., the subject at Spec; vP, valuing uninterpretable features on both sides. For illustration, let us study the VS sentences (13) and (14) from Saudi Arabic and Standard Arabic, respectively.

Table

13)	ʃa:f-u:	ʔil-ʕia:l	ʔil-film
	watched.3rd.PL.M	the-boys	the-film
	“The boys watched the movie.”

Table

14)	qaraʔa	l-ʔawlaad-u	l-qiṣat-a
	read.3^rd.SG.M	the-boys-Nom	the-story-Acc
	“The boys read the story.”

In (13) from Saudi Arabic, the verb ʃa:f-u: “watched.3^rd.Plural.Masculine” agrees with the postverbal subject ʔil-ʕia:l “the-boys” in all φ-features (person, gender, and number). By contrast, in (14) from Standard Arabic, the verb qaraʔa “watched.3^rd.Plural.Masculine” appears to agree with the postverbal subject l-ʔawlaad-u “the-boys” in person and gender features only. However, the VS derivations, (13) from Saudi Arabic and (14) from Standard Arabic, can be proceeded in a unified way and therefore represented as in (15a) and (15b), respectively.

short-legend15)

The VS derivation in (15a) and (15b) is advanced as follows. The head T, via FI, probes down for a matching goal within its c-commanding domain and finds the subject at Spec; vP. Under the probe-goal matching, the uninterpretable features on the verb (technically the probe T) get valued as third person, plural and masculine, i.e., it fully agrees with the subject at Spec; vP. Meanwhile, the Case is valued on the subject. The verb needs to raise to T so that temporal features of the verb, specifically tense, will get valued, thus deriving the correct word order (VS). The mere difference, I claim, is that the number feature agreement in Standard Arabic is spelled out periphrastically rather than affixially as it is the case in Saudi Arabic. Therefore, the peculiarity of the VS order of Standard Arabic lies in the way of spelling out the number feature of the verb at PF. This line of reasoning follows from two respects. First, the number feature (the feature that disappears in the VS of Standard Arabic, particularly with lexical subjects) is manifested in both word orders (VS and SV) of Arabic dialects. It indicates that the (morpho)phonological blocking of the number feature agreement in Standard Arabic, namely in lexical subjects of the VS order, is inapplicable in Arabic dialects. Thus, the loss of number feature agreement in the VS of Standard Arabic needs to be viewed as a PF operation. Second, the argument for the effect of PF operations on the absence of the number feature agreement is not novel. Benmamoun (2000) and Benmamoun & Lorimor (2006) propose that the subject and the verb undergo a PF merger in the VS order but not in the SV order; the number feature is spelled out affixially in the SV order but periphrastically in the VS order. Spelling out the number feature periphrastically rather than affixially is attributed to the fact that the subject serves as an exponent for realizing the number feature, making the presence of the number affix redundant. That is to say, when the syntactic derivation of full agreement reaches PF component for spelling out, a morphological merger takes place between the subject and the verb, leading the number feature agreement to appear periphrastically as argued by Benmamoun (2000). Due to this merger, the number feature “Plural” agreement on the verb disappears and is replaced by the default number feature “Singular”. Again, this merger option is a specific property of Standard Arabic. It is not a possible option for Arabic dialects, e.g., Saudi Arabic, which manifest all features, including the number, as an affix on the verb with the relevant order.

3.2. Full agreement in SV derivation: C→T_{φ-Features, EF}

In the derivation of SV sentences, the proxy head T inherits, from the phase head C, the Edge Feature along with uninterpretable φ-features. The probe T, via FI, establishes an Agree relation with a matching Goal, i.e., the subject at Spec; vP, thus valuing uninterpretable features on both sides. To satisfy the Edge Feature on the probe T, the subject is raised from to Spec; vP to Spec; TP. For illustration, let us study the examples (16) and (17) from Saudi Arabic and Standard Arabic, respectively.

Table

16)	ʔil-ʕia:l	ʃa:f-u:	ʔil-film
	the-boys	watched.3rd.PL.M	the-film
	“The boys watched the movie.”

Table

17)	ʔal-ʔawlaad-u	qaraʔ-u	l-qiṣat-a
	the-boys-Nom	read.3^rd.PL.M	the-story-Acc
	“The boys read the story.”

In (16), the verb ʃa:f-u: “watched.3^rd.Plural.Masculine” agrees with the preverbal subject ʔil-ʕia:l “the-boys” in all φ-features (person, gender, and number). Likewise, the verb qaraʔa “watched.3^rd.Plural.Masculine” in (17) fully agrees with the preverbal subject l-ʔawlaad-u “the-boys”. The two sentences of VS order, (16) from Saudi Arabic and (17) from Standard Arabic, can be similarly structured as in (18a) and (18b), respectively.

short-legend18)

In structures (18a-b), the SV derivation in both dialectal and standard varieties of Arabic is very similar. The probe T, via FI, establishes an Agree relation with the subject at Spec; vP in a C-commanding domain. The probe T has the Edge Feature [EF] together with φ-features by virtue of feature inheritance. By a matching relation between the head T and the subject, the uninterpretable features on the verb (technically T) get valued as “3^rd Person, Plural and Masculine” (full agreement morphology) and at the same time the Case is valued on the subject. The subject is raised to Spec; TP to satisfy the Edge Feature requirement of the probe T, thus deriving the SV order. In the SV derivations of Standard Arabic and Saudi Arabic, the number feature agreement, like other φ-features, is spelled out affixially on the verb at PF component.

4. Further consequences

The proposed account can respond to the debating question of whether the noun phrase in the SVO order of Standard Arabic is a real subject or a topic. Ouhalla (1991, 1994) and Plunkett (1993), following the traditional view of Arab grammarians (Ibn Hisham, 1964; Alafghani, 1974; Alghalayyini, 1974), exclude the possibility of having the subject in a preverbal position, i.e., it is a topic or focus. By contrast, Mohammad (2000), Fassi-Fehri (1993, 2005), Shlonsky (1997), Aoun et al. (1994), and Demirdache (1991) view that the preverbal noun phrases in the SV order as real subjects. Technically, if the preverbal noun phrase is a subject, then it should occupy the specifier position of TP. If it is a topicalized element, it should occupy the specifier position of C. The proposed account predicts the two possible scenarios. In both contexts, the subject must be generated at Spec; vP, a position where it can be accessed by the probe T for feature valuation and Case-wise property. In one scenario,³ the Edge feature on the probe T triggers the movement of the noun phrase from Spec; vP to Spec; TP. In the other scenario, the informational structure requirement, i.e., topic/foci, attracts the subject to its edge, triggering a further movement from Spec; TP to Spec; CP for topicalization/focus purposes.

5. Concluding remarks

The paper has argued that agreement-word order asymmetries in Arabic should be boiled down as modern dialects of Arabic always show full agreement between the subject and the verb irrespective of word order they may use. In addition, Standard Arabic in the VS order can display full agreement features between the verb and the subject when the latter is a pronoun, thus reducing agreement intricacies of Arabic into one paradigm, i.e., the full agreement features in the narrow syntax. The subject-verb agreement in standard and dialectal versions of Arabic takes place under an Agree relation between the probe T, via FI, and the subject generated at Spec; vP, where all φ-features of T get valued and at the same time the Case is valued on the subject. The lack/blocking of the number feature agreement morphology on the verb in the VS order of Standard Arabic is attributed to the existence of a morphological merger at the PF component, i.e., the realization of number feature affix on the verb was phonetically blocked. The paper has also proposed that deriving a possible word order relies on the presence or absence of the EF on the probe T, hence yielding SV and VS orders, respectively.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

The author has no funding to report.

Notes on contributors

Mustafa Ahmed Al-humari

Mustafa Ahmed Al-humari is an Assistant Professor of Linguistics in the Department of Languages and Translation, Faculty of Science and Arts-Rafha, Northern Border University, Saudi Arabia. His primary research area is formal syntax. He works on issues in Arabic morphosyntax, comparative syntax, and syntax-semantics interface. He has developed an inclination towards investigating language phenomena related to the syntax-pragmatics interface and language acquisition.

Notes

1. Licensing and Identification Conditions of pro:Licensing Condition: pro is Case-marked by X°_y, where y is parametrized.Identification Condition: pro inherits the ɸ-feature values of X°_y (if it has ɸ-features; if not, pro gets a default interpretation, typically arb). For more details, see Rizzi (1986).

2. I essentially adopt Koopman and Sportiche’s (1991) VP-Internal Subject Hypothesis that the subject of a finite sentence is generated within VP at D-Structure and then it moves to the specifier position of IP at S-Structure, thus deriving SVO word order.

3. The anonymous reviewer mentions a critical point that SVO order in Arabic remains a possible strategy of word order in the embedded context where C is normally filled with ʔinna where the subject appears to the right of C and not to its left. For illustration, see examples (i) and (ii).

Table

i)	qaal-uu	ʔinna	l-ʔawlaad-u		saafar-uu
	said-3^rd.Pl.M	that	the-children-Acc		traveled.3^rd.SG.M
	“They said that the children traveled.”
ii)	qaal-uu	ʔinna-hum			saafar-uu
	said-3^rd.Pl.M	that-them			traveled.3^rd.SG.M
	“They said that they traveled.”