24wa-raw [⌂ EvenYrbk]

Rami Alwardat

1.

Object Marking in Jordanian Arabic

Introduction

Object agreement is typically analysed as the realisation of phi features on v (Chomsky 2000, 2001), while clitic doubling is often seen as the movement of a D head into a verbal inflectional complex (Nevins 2011). Although these phenomena are distinct in theory, they can be challenging to differentiate in practice. This paper examines this issue in Jordanian Arabic (JA), focusing on the object marker morpheme. The object marker attaches to verbs and matches the phi features of an internal argument. For instance, in (1), -u is an object marker referring to the third person masculine singular direct object Sami. In another example, (2) -hə refers to Samia. In (3) -hum refers to l-wlaad ‘the boys’ demonstrating variation in the object marker’s form based on person, number and gender. These enclitics appear at the left edge of the verb. Depending on the structure, the object may be represented solely by an enclitic or by both an enclitic and a noun or pronoun. The omission of the freestanding pronoun or the noun (emphatic content) does not affect the utterance’s grammaticality and meaning. 1)

ʃuf-t-u Sami saw-1SG-3SGM Sami ‘I saw him, Sami.’

2)

ʃuf-t-hə Samia saw-1SG-3SGF Samia ‘I saw her, Samia.’

3)

ʃuf-t-hum l-wlaad saw-1SG-3PLM the-boys ‘I saw them, the boys’

4)

ʃuf-t-u saw-1SG-3SGM ‘I saw him, he.’

5)

ʃuf-t-hə hɪya saw-ISG-3SGF she ‘I saw her, she.’

huwa he

Alwardat, Object Marking in Jordanian Arabic 15

6)

ʃuf-t-hum humma saw-1SG-3PLM they ‘I saw them, they.’

When the direct object is nominal (emphatic content), leaving out the enclitic from the verbal complex does not affect the grammaticality of the construction. If the direct object (emphatic content) is pronominal, leaving out the enclitic renders the construction ungrammatical. 7)

ʃuf-t Sami saw-1SG Sami ‘I saw Sami.’

8)

ʃuf-t

huwa saw-1SG he Intended ‘I saw him.’

This paper investigates the status of object markers: whether they are reflexes of agreement (morphological realisations of phi-features on a functional head) or doubled clitics (pronoun-like morphemes associated with the direct object that moves and attaches to the verb). The paper concludes that, by examining predictions from current theories of agreement and clitic doubling against JA data, object markers are best classified as doubled clitics.

2.

Literature review

The JA object marker puzzle involves distinguishing between agreement morphemes and doubled clitics in languages with verb morphemes that vary with arguments. The boundaries between agreement and clitic doubling are often unclear due to undefined terms or theories. Canonical properties—like agreement being typically obligatory and clitic doubling often optional—help differentiate the two. The paper adopts specific theories of agreement and clitic doubling to assess whether the object marker aligns with these predictions, aiming to clarify the distinctions between them. This paper uses a Minimalist framework to explain agreement via the Agree relation, following Chomsky (2000, 2001, 2004). A functional head, like T (for subject agreement) or v (for object agreement), acts as a probe with unvalued phi-features. It searches its c-command domain for a DP with valued phi-features, known as the goal. Upon finding the goal, the probe and DP enter the Agree relation. The DP values the probe’s phi-features, and the probe assigns Case to the DP (nominative for T, accusative for v). The phi-features on the functional head manifest as agreement markers at PF. The Agree relation requires specific conditions: the goal must be within the c-command domain of

Alwardat, Object Marking in Jordanian Arabic 16

the probe, and both the probe and goal must be “active,” meaning both must have unvalued features. The probe’s unvalued features are phi-features, while the goal DP has an unvalued Case feature. These conditions are crucial for the Agree relation to occur. However, this condition may lead to incorrect predictions regarding the behaviour of JA object markers, as will be discussed later in the analysis. 9)

Agreement markers are typically morphophonologically dependent and can be attached as affixes or clitics to a host. While agreement morphemes may function as clitics—elements less dependent than affixes but more so than words—they differ from doubled clitics, which arise through a separate process. There are two primary analyses of clitic doubling: one treats the clitic as a non-canonical agreement marker (see Sportiche 1996), while the other views it as a morpheme that has moved from within the determiner phrase (DP) into the verbal complex (see Uriagereka 1995). Some research combines both analyses based on clitic type (see Ormazabal and Romero 2010). In Minimalism, a movement approach is favoured, as it better fits the framework and aligns with evidence that doubled clitics are of category D. This approach, which posits the clitic as a D head with a null noun phrase (NP) complement and the doubled DP as its specifier, will be adopted to maintain consistency with Minimalist assumptions about grammar and to explain specific object marker properties. Though details of this movement approach will not be central in the discussion, it underpins the analysis of clitic doubling in this context.

Alwardat, Object Marking in Jordanian Arabic 17

10)

Big DP Hypothesis: (Uriagereka 1995:81, (2a))

In Uriagereka (1995), clitics are analysed as syntactic heads (D) that raise to a functional head (FP) near the verb rather than being tied to Case assignment or phi-feature valuation. This distinguishes clitic doubling from agreement, where agreement markers are phi-features on heads like T or v. Clitic doubling doesn’t involve the exact mechanisms of agreement, particularly in the context of JA. The paper argues that the behaviour of object markers aligns with clitichood rather than agreement, highlighting the fundamental differences between the two phenomena. Paparounas and Salzmann (2024) explore clitic doubling in conjunction constructions in Modern Greek, challenging movement-based approaches to clitic doubling. They argue that movement-based theories incorrectly predict that the clitic doubling seen in examples such as (11) should violate the Coordinate Structure Constraint. Instead, their evidence supports an Agreebased analysis, where doubled DPs remain in their base positions. 11)

{Se 2SG.ACC [esena you.ACC

/sas /*tin} iða 2PL.ACC 3SG.F.ACC see.PST.1SG ke ti Maria] sto parko and the.ACC Mary.ACC in.the.park

The authors reanalyse the properties of Greek clitic doubling phenomena, such as the alleviation of Weak Crossover effects and suspension of intervention effects, through Agree and information structure (givenness, topicality). They propose that clitic doubling can be explained by a licensing approach, similar to differential object marking, rather than movement. Additionally, they address morphological challenges to Agree-based accounts, specifically the syncretism between determiners and clitics and tense invariance. They argue that syncretism poses no more significant challenge to Agree-based theories than movement-based ones and that tense invariance is not a reliable indicator to distinguish agreement from clitic doubling.

Alwardat, Object Marking in Jordanian Arabic 18

Jarrah et al. (2024) present evidence from Modern Standard Arabic and Jordanian Arabic showing that when a probe undergoes φ-Agree with a pro (a phonologically null element), it results in a φ-inflection on the probe. This is explained by the “Agree Identification Condition,” which requires a phonologically null goal to be identified through a corresponding φ-inflection on its probe. This analysis accounts for why φ-inflections in Arabic do not appear on verbs or prepositions when these elements φ-agree with an overt DP. The article further demonstrates that this condition does not apply when the Agree relation occurs with an unpronounced goal that is part of a movement chain. If one link of the chain, such as the higher copy, is overt, then no φinflection appears on the probe. This implies that when a probe φ-agrees with a goal that is part of a movement chain, the Agree relation involves the entire chain rather than just the individual links.

3.

Morphophonology

The object marker’s morphophonological form suggests it functions as a doubled clitic rather than an agreement morpheme. Agreement morphemes typically reflect phi-features (like person, number, or gender) on functional heads, such as T (for subject agreement) and transitive v (for object agreement). The form of an agreement morpheme can vary based on additional features of the functional head, such as tense. For example, in English, the agreement morpheme varies with tense: it is null in past tense verbs but manifests as “-s” in the 3rd person singular present tense. This variation aligns with the idea that agreement morphemes are contextually dependent on the features of the functional heads they attach to. The object marker, by contrast, does not exhibit this kind of variation, indicating it is likely a clitic, which is more independent of the functional head’s features. In JA, affixes cross-referencing the nominative subject consistently maintain a linear order closer to the verbal stem than object markers. Morphologically, inflectional subject markers are distinguished from object markers by their placement as suffixes in the past tense (perfect) (12) and as prefixes (13) or circumfixes (14) in the present tense (imperfect). Object markers consistently suffix verbs (enclitics), indicating the object’s presence (15). 12)

ʔɪl-bɪnɪt leʕb-ət the-girl played-3SGF ‘The girl played.’

(perfect)

Alwardat, Object Marking in Jordanian Arabic 19

13)

14)

15)

ʔɪl-bɪnɪt btɪ-lʕab the-girl 3SGF-play ‘The girl plays.’

(imperfect)

ʔɪl-banaat bɪ-lʕab-ɪn the-girls play-3PLF ‘The girls play.’

(imperfect)

Samia zaar-ət-nə Samia visited-3SGF-1PL ‘Samia visited us yesterday.’

ʔɪmbareħ yesterday (perfect)

The perfect and imperfect agreement suffixes differ in morphophonological form and position. In the perfective aspect, the agreement morpheme is suffixed to the verb, while it is prefixed in the imperfective aspect. However, some imperfect forms exhibit both prefixal and suffixal agreement. A spectrum of subject marking possibilities is highly sensitive to tense. The property of tense sensitivity cannot be found in object marking. Object markers are always suffixes in JA, and the form of the object marker itself is not sensitive to tense. Sensitivity to tense is known to be a property of inflection, whereas insensitivity to tense has been presented as a key diagnostic for clitichood (see Nevins 2011). In (15), the linear ordering of subject and object marking (with object marking peripheral to subject marking) is an empirical indication that object marking is not inflection. The alternative is that it is a clitic. If object markers were inflections, the Mirror Principle (Baker 1985) would lead to the expectation that the object markers would have to be closer to the stem than the subject markers since object marking is lower in the tree than the subject marking. Jarrah et al.’s. (2024) account violates the Mirror Principle, as (15) indicates. The Minimalist theory of agreement allows for variation in agreement morphemes across different verb forms (e.g., present tense, past tense, imperfect aspect), observed cross-linguistically and in JA. However, the object marker remains completely invariant across all verb forms, unaffected by properties of the transitive v or any aspect of the verb form. Despite being part of the same Morphological Word as subject agreement, the object marker’s form does not change based on aspect or other components. Instead, it varies solely according to the phi-features of the associated argument. This invariance suggests that the object marker is a clitic, distinct from affixes in JA, highlighting a clear morphophonological difference between affixes and clitics. The accompanying table further supports this analysis by presenting the full paradigm of object markers in the context of other indicators of agreement, showing their consistent form across different contexts.

Alwardat, Object Marking in Jordanian Arabic 20

Table 1: Strong pronouns, subject markers and enclitics in JA

1S

Nominative Strong Pronouns Ɂana/ Ɂanɪ

Enclitics Accusative clitic

Subject Markers/Affixes -t

Genitive clitic

-nɪ

-ɪ

1P

ʔɪħnə

-nə

2SM

Ɂɪntə

-t

-ək

2SF

Ɂɪntɪ

-tɪ

-kɪ -ɪʧ

-ɪk -ɪʧ

2PM

Ɂɪntu

-tu

-ku

2PF

Ɂɪntɪn

-tɪn

-kɪn -ʧɪn

3SM

huwə

-Ø

-u

3SF

hɪyə

-ət

-hə

3PM

hummə

-u

-hum

3PF

hɪnneh

-ɪn

-hɪn

The object marker’s morphophonological behaviour aligns more with pronominals than with a bundle of phi features on a verbal functional head. This supports a ‘big DP’ treatment of the object marker as a clitic, where the clitic functions as a D head, as predicted by Anagnostopoulou’s 2004 analysis of Greek clitics, which sees them as the realisation of a moved [D] feature of a DP. Moreover, Table 1 shows that the object marker and pronominal possessor are formally identical in eight out of ten paradigm slots. Consider the following examples: 16)

madrasat nə school 1PL ‘our school’

17)

beɪt hum house 3PLM ‘their house’

An agreement approach could explain the syncretism in object markers if pronominal possessors are treated as possessor agreement, making object and possessor agreement syncretic. However, this is doubtful since pronominal

Alwardat, Object Marking in Jordanian Arabic 21

possessors cannot co-occur with overt possessors, unlike in Hungarian, where such possessor agreement is possible (Szabolcsi 1994). 18)

beɪt-hə lə-Samia house-3SGF to-Samia ‘Samia’s house’

19)

lə-beɪt-hə

of-house-3SGF ‘Samia’s house’

Samia Samia

If pronominal possessors are analysed as D heads, this syncretism is easily explained, as both would represent a D with phi-features. The strong formal parallels suggest a significant historical connection. This similarity ties the object marker more closely to the nominal system than the verbal system, emphasising its morphophonological alignment with nouns rather than verbs. The obligatory absence of the determiner ʔɪl ‘the’ attaching to the noun beɪt in (20) suggests that the DPs are headed by a phonetically null determiner, Dgen, assigning genitive case, preventing the structure from being headed by two determiners simultaneously. 20)

ʔɪl-beɪt-hə

The-house-3SGF ‘Samia’s house’

lə-Samia to-Samia

In summary, the object marker’s tense invariance contradicts its classification as an agreement morpheme but aligns with a clitichood account. Additionally, the object marker shows formal parallels to pronominal possessors, suggesting a connection to the nominal morphology system, further supporting its identification as a doubled clitic.

4.

Distribution

The object marker in JA is typically optional, but it emphasises the referenced argument when used. In Arabic, emphasis is employed to help the listener avoid ambiguity. It is only allowed for definite, specific arguments, though the precise rules governing this are not yet fully understood. 21)

ʃuf-t-u lɪ-mʕəlɪm saw-1SG-3SGM the-teacher ‘I saw him, the teacher.’

Alwardat, Object Marking in Jordanian Arabic 22

22)

ʃuf-t lɪ-mʕəlɪm saw-1SG the-teacher ‘I saw the teacher.’

23)

ʃuf-t-u

mʕəlɪm saw-1SG-3SGM Teacher Intended, ‘I saw him the teacher.’

24)

ʃuf-t mʕəlɪm saw-1SG Teacher ‘I saw a teacher.’

In (21), the definite direct object “lɪ-mʕəlɪm” ‘the teacher’ allows optional attachment of an object marker to the verb. In (23), the indefinite direct object “mʕəlɪm” ‘a teacher’ cannot take an object marker, making it ungrammatical. When an argument is pro-dropped, the object marker becomes obligatory. Consider (25) and (26). 25)

ʔakal u ate 3PLM ‘They ate.’

26)

ʔakal u ate 3PLM ‘They ate it.’

hə 3SGF

In (25), the absence of an object marker renders the sentence grammatical only when no specific direct object is implied (Specificity is a property of clitic doubling constructions: non-specific objects typically can’t be doubled by a clitic) (see Kallulli & Tasmowski 2008), similar to the English equivalent. However, if the sentence suggests a pro-dropped direct object, an object marker becomes necessary, as shown in (26). This pattern resembles clitic doubling in River Plate Spanish, where clitic doubling is optional for full DP direct objects, mandatory for dropped pronominal objects, and influenced by object specificity (see Jaeggli 1982). 27)

Le duele la cabeza her-dat hurts the head ‘Mafalda has a headache’

a Mafalda to M

28)

Le rompi la pata a her-dat I-broke the leg to ‘I broke the table’s leg.’

29)

Le entreguê la carta a él ‘I delivered the letter to him’

la Mesa. the Table

(Jaeggli 1982:13)

Alwardat, Object Marking in Jordanian Arabic 23

This similarity supports the idea that the object marker may be a double clitic in JA since agreement typically isn’t optional, doesn’t usually carry semantic effects, and isn’t generally conditioned by the controller’s features in this language. A relevant observation is that in JA, the object marker can only establish a relation with a definite or specific DP. If the argument is indefinite, clitic doubling becomes ungrammatical, indicating that the object marker may behave more like a clitic than an agreement marker, depending on the DP’s definiteness and specificity. JA object markers may represent a rare type of agreement, but the likelihood of all these non-canonical features appearing together in one language seems low. A more precise argument can be made using Preminger’s (2009) diagnostic distinguishing between agreement and clitic doubling. Agreement involves valuing preexisting unvalued features on a functional head, while clitic doubling generates or merges a pronoun-like morpheme. These processes yield different outcomes when they fail, allowing for a more precise analysis of the underlying mechanism. The diagnostic, outlined in (30), uses these distinctions to better understand the phenomena. 30)

Given a scenario where the relation R between a morpheme M and the corresponding full noun phrase X is broken but the result is still a grammatical utterance the proposed diagnostic supplies a conclusion about R as follows: a. M shows up with default phi-features (rather than the features of X) → R is Agree; b. M disappears entirely → R is clitic doubling. (Preminger, 2009: 623)

Preminger’s diagnostic differentiates between agreement and clitic doubling by examining what happens when these relations are disrupted. In agreement, disruption (e.g., by defective intervention) can result in default phi-features on the probe, as seen in Icelandic. In clitic doubling, if locality conditions are violated (e.g., the clausemate relation), the structure may still be grammatical, but the clitic doesn’t appear, as in Basque. Applying this to JA is challenging since the locality conditions of object markers are still being studied. If a default object marker is grammatical, it indicates object agreement. If the absence of an object marker is grammatical, it suggests clitic doubling. However, a default object marker (third-person masculine singular) is ungrammatical, supporting the clitic doubling interpretation. Consider (21)-(26) above and (31) below. 31)

ʃuf-t-u

lɪ-mʕəlɪm-eh saw-1SG-3SGM the-teacher-3SGF Intended’ I saw him, the female teacher.’

Alwardat, Object Marking in Jordanian Arabic 24

5.

Case

In Minimalist theory (Chomsky 2000, 2001, 2004), object agreement is tightly linked to accusative Case. When a verb’s v head agrees with a DP, it assigns accusative Case to that DP. DPs with already valued Case features can’t enter further Agree relations, meaning only DPs receiving accusative Case should participate in object agreement. This is not true for JA object markers, where the distribution of object marking is not tied one-to-one to the distribution of accusative case assignment. Thus, an accusative Case isn’t necessary for a DP to relate to an object marker. For instance, indirect objects in JA, which receive Case via the preposition lə- (to), can still be referenced by an object marker. In example (32), the object marker -hə refers to the female indirect object lɪmʕalm-eh ‘the teacher’, not to the masculine direct object lɪ-ktaab ‘book’. This shows that JA object markers don’t strictly follow the predicted link between object agreement and accusative Case. An object marker in JA can reference an argument without assigning accusative Case, challenging the predictions of an agreement-based analysis as Jarrah et al (2024) account. 32)

Sami ʔaʕtˤə-∅-hə lɪ-ktaab lə-lɪ-mʕalm-eh Sami gave-3SGM-3SGF the-book to-the-teacher-3SF ‘Sami gave her the book to the teacher.’

The data in (32) challenge an agreement analysis. The DP lɪ-mʕalm-eh ‘the teacher’ already has its Case feature valued as dative by the preposition lə’to’, making it inactive in Minimalist terms. Consequently, the v probe shouldn’t be able to enter an Agree relation with this inactive DP. However, contrary to this prediction, the object marker -hə on the verb does reference lɪ-mʕalm-eh. This suggests that the principles of object agreement do not strictly govern object markers in JA. Notice that “dative case” is not morphologically visible on nominal arguments in JA. Morphological “dative case” is visible in pronominal arguments. Accordingly, I argue that JA has no biunique link between object marking and accusative. Consider (33): 33)

Sami ʔaʕtˤə-∅-nɪ lɪ-ktaab Sami gave-3SGM-1SG the-book ‘Sami gave me the book to ME.’

ʔɪl1-ɪ to-1SG

In some instances, referencing an argument with an accusative case using an object marker is ungrammatical. For example, in (34), the reflexive direct object ħaal-hə ‘herself’ has accusative case but cannot be referred to by an object 1

Another phonological variant of the preposition lə.

Alwardat, Object Marking in Jordanian Arabic 25

marker on the verb. The use of an object marker on the verb renders the sentence ungrammatical. This indicates that accusative case is neither necessary nor sufficient for referencing by an object marker, making it unlikely that the object marker functions as a form of object agreement. 34)

Samia ʃaaf-ət-hə

ħaal-hə bɪ-lɪ-mraayah Samia saw-3SGF-3SGF self-3SGF in-the- mirror Intended, ‘Samia saw herself in the mirror.’

In summary, an agreement analysis inaccurately predicts the behaviour of object markers with accusative case. Conversely, a clitic doubling analysis does not link doubled clitics tightly to accusative case, so it aligns well with the observed data.

6.

Binding

The object marker in JA influences binding relationships. While a subject can bind a possessive pronoun in the direct object, backward pronominalization between subjects and objects is nearly ungrammatical in JA, as shown in the following examples: 35)

Samiai ʃaaf-ət tˤullaab-həi Samia saw-3SGF students-POSS.3SF ‘Samia saw her students.’

36)

tˤullaab-həi

ʃaaf-u students-POSS.3SGF saw-3PLM Intended: ‘Her students saw Samia.’

Samiai Samia

The construction in (36) can be fixed when the object is referenced by an object marker, as in (37). 37)

tˤullaab-həi ʃaaf-u-həi students-POSS.3SGF saw-3SGF ‘Her students saw her, Samia.’

lə-Samiai to-Samia

The object marker facilitates the binding of the object into the subject, similar to how clitic doubling influences binding relationships, including backward pronominalization, as observed in Greek (Anagnostopoulou 2003: 207208). Clitic doubling involves the movement of a pronoun-like element closely tied to binding relations. In contrast, minimalist agreement theory suggests that agreement markers, bundles of uninterpretable phi features, should not affect binding since they cannot refer to or alter binding relations (Rezac 2010).

Alwardat, Object Marking in Jordanian Arabic 26

38)

a. b.

Some linguistsi/Theyi seem to them(selvesi) to have been given good job offers. There seem to them(*selvesi)/*each otheri to have been some linguistsi given good offers. (Rezac 2010: 497)

Therefore, the fact that the object marker enables backward pronominalization suggests it functions as a double clitic rather than merely an agreement marker, highlighting its role in binding relationships.

7.

Grammaticalisation

Through phonological changes and grammaticalisation, freestanding pronouns in Arabic dialects became object markers. Case vowels disappeared in Arabic varieties. In MSA, case is typically indicated by the final vowel of a word. This vowel can be u (nominative case for the subject), ə (accusative case for the object), or ɪ (genitive case for possession). Freestanding pronouns lost their original function and started marking verbal complements in specific contexts. Eventually, the nominative independent pronouns expanded to all syntactic environments, illustrating a shift in their role and distribution across the language’s structure. Lipiński (2001: 308) states that “the independent personal pronoun of Proto-Semitic most likely possessed at least one non-subject case”. Arabic syllable structures primarily consist of CV, CVV, and CVC, with every syllable beginning with a consonant (Watson 2002, 56). Building on this foundational structure, the transition of strong pronouns into clitics is discussed, drawing on the explanation by JapenSarage and Kasiyarno (2015). The pronoun naħnu “we” illustrates this process. It comprises two syllables, naħ and nu, with stress on the first syllable. When the stress and second syllable are lost, the remaining element, na, becomes an enclitic. This clitic attaches to the end of a host word, as seen in baytu-nə, “our house”. Similarly, the first-person singular pronoun ʔənə: “I” contains two syllables: ʔə and nə. With the deletion of the second syllable, the remaining syllable, ʔə, transforms into a clitic that attaches to its host. As clitics cannot bear the stress, the glottal stop /ʔ/ weakens, merging with the vowel /ə/ to produce /ɪ/. The resulting weak form, -ɪ, serves as the first-person singular clitic, as in bayt-ɪ: “my house”. In addition to attaching to nouns, clitics can attach to prepositions, as in l-ɪ, “I own”. When clitics exhibit word-like properties, they can connect to various word categories, including verbs, nouns, prepositions, and quantifiers. The second-person singular independent pronouns ʔəntə (masculine) and ʔəntɪ (feminine) undergo a similar reduction. Both have two syllables: ʔən and tə (for masculine) and ʔən and tɪ (for feminine). The first syllable is deleted,

Alwardat, Object Marking in Jordanian Arabic 27

leaving tə and tɪ. Over time, the consonant /t/ weakens, becoming /k/. This phonological change results in the clitics -kə for the second-person masculine singular, as in baytu-kə “your house”, and -kɪ for the second-person feminine singular, as in baytu-kɪ “your house”. The syllabic structures of the third-person singular pronouns huwə (masculine) and hɪyə (feminine) are hu-wə and hɪ-yə. The second syllables are deleted in these cases, leaving hu and hɪ. The feminine clitic—hɪ then changes into -hə. These clitics, -hu and -hə, serve as third-person singular markers, as in baytu-hu “his house” and baytu-ha “her house”. It is essential to note the parallel between clitic formation and affixation. Both are derived from independent pronouns, and neither can function independently, requiring attachment to a nearby word. Clitics are smaller than their independent pronoun counterparts, while affixes are often similar in size or smaller than clitics. Pronouns lose independence over time. Initially, a pronoun loses a syllable or stress, becoming a clitic. In the final stage of grammaticalisation, the clitic evolves into an affix, marking subject-verb agreement features. Notably, there are similarities in pronominal clitics between JA and Modern Standard Arabic (MSA). For instance, the first-person singular clitic -ɪ/nɪ and plural -nə remain consistent across both varieties. Likewise, the third-person plural clitic -hum is identical, whereas the second-person plural clitic -kum changed into -ku. However, changes occur with pronouns ending in short vowels. For example, the second-person masculine singular clitic -kə becomes -ək, while the second-person feminine singular clitic -kɪ changes to -ɪk. Regional and social factors influence these shifts. Urban speakers often use -ɪk, while rural speakers, particularly elders, may use -ɪtʃ. Similarly, the third-person singular masculine clitic -hu shifts to -u, while the feminine form ha remains unchanged in both varieties. Changes also affect the second- and third-person feminine plural clitic -hunna, which evolves into -hɪn. Evidence of grammaticalisation in freestanding pronouns is seen in Arabic Varieties like Lebanese and Coastal Syrian dialects, which retain independent object pronouns. This archaic feature, preserved in these vernaculars, appears with ditransitive verbs such as ʔaʕtˁə “give” and ʤaab “bring”, where an independent pronoun can stand as a pronominal direct object as in the following examples: 39)

ʔaʕtˤeɪ-t-ak huwa gave-1SG-2SGM 3SGM.he ‘I gave you it.’

40)

ʤɪb-t l-ha brought-1SG for-3SGF ‘I brought her them.’

hɪnne 3PLF.they

Alwardat, Object Marking in Jordanian Arabic 28

In the above examples, the freestanding pronoun or an object marker cliticised on the determiner ʔɪyya can fit in the object position. 41)

ʔaʕtˤeɪ-t-ak gave-1SG-2SGM ‘I gave you it.’

ʔɪyya-ha ʔɪyya-3SGM

These examples demonstrate that dependent object pronouns originate from independent ones through syllable reduction or phonological changes. Their ability to attach to various lexical categories, including verbs, nouns, prepositions, and quantifiers, confirms their status as clitics.

8.

Clitic doubling in JA

In JA, an object clitic can accompany a full DP associate at the right sentence periphery, showing agreement in gender, person, and number. The clitic pronoun may also appear without the full DP. Consider (42) and (43). 42)

zur -ɪt -hum visit-PAST -1PL -3PLM ‘I visited them, the boys.’

43)

zur -ɪt -hum visit-PAST -1P -3PM ‘I visited them, the boys.’

lə lə

lɪ-wlaad the-boys lɪ-wlaad the-boys

The use of lə- is optional when the subject is a first or second-person pronoun, as in (42) and (43), but it is a must when the subject is a third person. The optionality of lə- challenges Kayne’s Generalization in clitic doubling (Anagnostopoulou 2003:16). Anagnostopoulou (2003: 16) argues that clitic doubling typically requires a preposition before the object DP, but this variability suggests a deviation from Kayne’s rule. The use of lə- in JA is similar to that of Differential Object Marking in Romance; for example, Spanish exhibits DOM, where particular direct objects are marked with the prepositional a while others remain unmarked. In (44), a is DOM. This has been shown to originate from a strategy to disambiguate sentences with two 3rd person arguments: using a Pelement in front of one of the two DPs identifies that DP unambiguously as the object as in (45) and (46). 44)

lo vimos CL.3SG.M we.saw ‘We saw him/Juan’

a Juan to Juan

Alwardat, Object Marking in Jordanian Arabic 29

45)

Sami zaar

∅

Sami visited -3SGM ‘Sami visited her, Samia.’

-hə 3SGF

46)

Sami zaar -hə -∅ Sami visited -3SGM 3SGF ‘Sami visited her, Samia.’

Samia Samia lə lə

Samia Samia

It is worth mentioning that when an object marker does not accompany an object, it must be adjacent to the verb where an adverbial cannot separate it. (47) and (48) demonstrate this. 47)

Sami zaar Samia lɪ-sbuuʕ -∅ Sami visited -3SGM Samia week ‘Sami visited Samia last week.’

48)

Sami zaar

lɪ-sbuuʕ -∅ Sami visited -3SGM week ‘Sami visited her last week, Samia.’

ʔɪl-maðˤɪ Past

ʔɪl-maðˤɪ past

Samia Samia

The clitics, such as doubling in JA, might be compromised by the possibility of inserting an adverbial between the verb and the PP, consistent with PP dislocation. 49)

Sami zaar -hə lɪ-sbuuʕ -∅ Sami visited -3SGM -3SGF week ‘Sami visited her last week, Samia.’

50)

Sami zaar

hə

lɪ-sbuuʕ -∅ Sami visited -3SGM -3SGF week ‘Sami visited her last week, Samia.’

ʔɪl-maðˤɪ past ʔɪl-maðˤɪ past

lə lə

Samia Samia Samia Samia

In (49), the insertion of an adverbial between V+clitic and lə-Samia suggests that the DP is not an object. Contrastingly, (50) is ungrammatical with an adverbial between V and the DP, indicating that the clitic and DP can’t coexist within the same domain. Object clitics in JA never double full DP objects locally, indicating complementary distribution. Thus, object clitics are cliticized pronouns serving as arguments, not agreement forms (i.e. inflection).

9.

Conclusion

The fact that object marking is farther from the stem than subject marking indicates that object marking in JA is cliticisation. The tense invariance versus tense sensitivity of subject marking can also support this. This paper argues that object markers in JA are not agreement inflections but doubled clitics. The analysis is

Alwardat, Object Marking in Jordanian Arabic 30

based on the different predictions made by theories of agreement versus clitic doubling. Evidence is drawn from the morphophonological form of the object marker, its lack of interaction with Case, its distribution, and the grammaticalisation of the freestanding pronouns over time, all of which support the conclusion that the object marker functions as a doubled clitic. The paper distinguishes between agreement and clitic doubling and lays the groundwork for a detailed analysis of object markers in JA. While clitic doubling in JA initially appears similar to that in River Plate Spanish, JA’s clitic doubling is distinct in its ability to occur with various internal arguments, not just indirect objects. This sets it apart from the more familiar Romance cases of clitic doubling. Additionally, the obligatory use of the object marker in possessive sentences challenges previous research findings. Given the extensive research on clitic doubling, identifying JA object markers as doubled clitics could help refine understanding of this phenomenon across languages. Clitic doubling needs more elaboration in JA and due care within the linguistic theory. References Anagnostopoulou, Elena. 2003. The syntax of ditransitives: Evidence from clitics. Berlin: Mouton de Gruyter. Anagnostopoulou, Elena. 2004. On clitics, feature movement and double object alternation. In Arthur Stepanov, Gisbert Fanselow, Ralf Vogel. (eds.). Minimality effects in syntax. Berlin: Mouton. 15–36. DOI: https://doi.org/10.1515/9783110197365.15 Baker, Mark.1985. The Mirror Principle and Morphosyntactic Explanation. Linguistic Inquiry 16: 73-415. Chomsky, Noam. 2000. Minimalist inquiries, the framework. In Roger Martin, David Michaels, and Juan Uriagereka (eds.). Step by step: Essays on minimalist syntax in honor of Howard Lasnik. Cambridge: MIT Press. 89–155. Chomsky, Noam. 2001. Derivation by phase. In Michael Kenstowicz (ed.). Ken Hale: A life in language. Cambridge: MIT Press. 1–52. https://doi.org/10.7551/mitpress/4056.003.0004 Chomsky, Noam. 2004. Beyond explanatory adequacy. In Adriana Belletti (ed.). Structures and Beyond. New York: Oxford OUP. 104-131. https://doi.org/10.1093/oso/9780195171976.003.0004 Jaeggli, Osvaldo. 1982. Topics in Romance syntax. Dordrecht: Foris. https://doi.org/10.1515/9783112420225 Jarrah, Marwan, Mutasim Al‐Deaibes, and Yazeed Hammouri. 2024. ɸ‐Agree with silent goals and the theory of interface effects. Studia Linguistica 78(2): 254279. https://doi.org/10.1111/stul.12229 JapenSarage, M.A, and M. Kasiyarno. 2015. Arabic personal pronouns as word, clitic, and affix. International Journal on Studies in English Language and Literature: 33-41. Kallulli, Dalina and Tasmowski, Liliane. 2008. Introduction: Clitic doubling, core syntax and the interfaces. Clitic Doubling in the Balkan Languages, edited by Dalina Kallulli and Liliane Tasmowski, John Benjamins Publishing Company, pp. 1-32. https://doi.org/10.1075/la.130.03kal.

Alwardat, Object Marking in Jordanian Arabic 31 Lipiński, Edward. 2001. Semitic Languages: Outline of a Comparative Grammar. Leuven: Peeters. Nevins, Andrew. 2011. Multiple agree with clitics: Person complementarity vs omnivorous number. Natural Language & Linguistic Theory 29: 939–971. https://doi.org/10.1007/s11049-011-9150-4 Ormazabal, Javier, and Juan Romero. 2010. Object clitics and agreement. Ms., University of the Basque Country/HiTT and University de Extremadura/HiTT. Paparounas, L., Salzmann, M. 2024. First conjunct clitic doubling in Modern Greek: Evidence for Agree-based approaches to clitic doubling. Nat Lang Linguist Theory 42: 323–382 https://doi.org/10.1007/s11049-023-09585-2. Preminger, Omer. 2009. Breaking agreements: Distinguishing agreement and clitic doubling by their failures. Linguistic Inquiry 40: 619–666. https://doi.org/10.1162/ling.2009.40.4.619 Rezac, Milan. 2010. φ-Agree versus φ-feature movement: Evidence from floating quantifiers. Linguistic Inquiry 41: 496–508. https://doi.org/10.1162/LING_a_00007 Sportiche, Dominique. 1996. Clitic constructions. In Johan Rooryck and Laurie Zaring (eds.). Phrase structure and the lexicon. Dordrecht: Kluwer. 213–276. https://doi.org/10.1007/978-94-015-8617-7_9 Szabolcsi, Anna. 1994. The noun phrase. In Ferenc Kiefer and Katalin É. Kiss (eds.). The syntactic structure of Hungarian. Syntax and semantics 27. San Diego: Academic Press: 179–274. https://doi.org/10.1163/9789004373174_004 Uriagereka, Juan. 1995. Aspects of the syntax of clitic placement in western Romance. Linguistic Inquiry 26: 79–123. Watson, Janet CE. 2002. The phonology and morphology of Arabic. New York: Oxford University Press. https://doi.org/10.1093/oso/9780199257591.001.0001 Rami Alwardat Eötvös Loránd University ramiwardat2@student.elte.hu