Pronominalization and clitic doubling in Syrian and Omani Arabic

2.1 Pronominalization in double-object constructions in Damascus Arabic

In Damascus Arabic, a verb may only host one clitic pronoun suffix. It is not possible to cliticize both arguments of a double-object verb to the verb simultaneously. Which of the two objects can be cliticized to the verb in ditransitive constructions is grammatically conditioned in ways described below. We begin by describing two classes of double-object verb in Damascus Arabic and the basic pronominalization patterns found there. Then, we turn to clitic pronoun doubling in the two classes. The two classes of double-object verb differ in the way they mark the DP bearing the recipient thematic role. For both verb classes, the theme is morphologically unmarked, which we take to be an expression of accusative case. One class of verbs also assigns (unmarked) accusative to the recipient as well (like English and other languages, including Classical Arabic), while the other class assigns dative to the recipient (like German, Russian and other languages). Both verb types alternate with what we call a ‘prepositional’ frame in which the theme is accusative and the recipient occurs in a prepositional phrase. We begin with the double accusative class exemplified by the verb ʕaṭa ‘give’.

As in English, the two objects of ʕaṭa occur obligatorily in the order recipient > theme, as shown in (7a). We refer to this complement frame as the ‘double-object’ frame. In this order, the recipient may be pronominalized in the accusative clitic pronoun paradigm, shown in (7b). Also, in the passive, it is the recipient that raises to subject, as seen in (7c).

As in (standard) English, the two objects may not be reversed in the double-object frame, as (8a) shows. Furthermore, the theme may not appear as a clitic pronoun on the verb when a recipient DP occurs in the same clause, as (8b) illustrates. Nor may the theme be raised to subject in this case, as (8c) shows.

We take these facts to mean that the clitic pronoun is in a syntactic dependency with a gap in the position otherwise occupied by the non-clitic pronoun counterpart of the clitic pronoun, and that this dependency may not cross over the recipient, as schematized in (9). Consequently, the recipient itself may cliticize to the verb but the theme may not. The facts reviewed in Section 2 make clear that this is not a restriction on cliticization of the theme as such. The theme can very well cliticize to the verb in monotransitive constructions like that illustrated in (1b), which do not license a recipient argument. The facts reviewed above indicate that the recipient is an ‘intervener’ for cliticization of the theme to the verb.

(9)

The same can be said of raising of an object to subject in the passive. Raising to subject involves a dependency between the subject position (signified by ‘subj’ below) and a gap in a position that the subject would otherwise have occurred in. Like cliticization, this dependency is interrupted by an intervening DP, as schematized in (10).

(10)

In these examples, there is a correspondence between whether a DP can be cliticized to the verb and whether it can be promoted to subject in the passive. This correspondence is preserved in the alternative complement frame seen in (11a). Here, as in English, the theme directly follows the verb and the recipient occurs in a prepositional phrase, as (11a) illustrates. In this case, the theme may cliticize to the verb in the accusative paradigm, as shown in (11b), and raise to subject in the passive, as (11c) shows. The preposition that introduces the recipient argument in this frame is none other than la- ‘to’, the same preposition that marks the full DP double of a clitic pronoun in clitic pronoun doubling constructions. We refer to the alternative complement frames for ʕaṭa and similar verbs seen in (7a) and (11a) as the ‘double-object alternation’, and flesh out the nature of the syntactic relatedness of the two frames on the basis of clitic pronoun doubling patterns described in detail in Section 2.2. We refer to these frames as alternative expressions of the ‘double-object construction’ (but only the former in (7a) is the ‘double-object frame’).

Damascus Arabic has a paradigm of object clitic pronouns that share the l- onset of la-. This paradigm reflects pronominalization of a dative DP but not a prepositional phrase. The prepositional phrase in (11a) cannot be pronominalized in the l- paradigm, as shown in (12a). On the other hand though, this is not expected to be possible for independent reasons: the fact that the theme intervenes between it and the verb, its potential morphological host, could be expected to have the same blocking effect as the recipient does for the theme in (8b). However, ʕaṭa still does not allow cliticization of the la-phrase to the verb when the theme is moved out of the way by raising to subject in the passive, as shown in (12b). It appears that ʕaṭa does not license dative case; its la-phrase argument in (11) is a PP that cannot cliticize to the verb.

The second class of double-object verbs in Damascus Arabic is exemplified by baʕat ‘send’. Its recipient argument is invariantly marked by the particle la-.

Hallman (2018: 242–246) discusses reasons to believe that the la- that occurs in (13) is not the preposition la- seen in (11), but rather a case marker, so that la-sāra is a DP in (13) while it is a PP in (11). For one, in the case of ʕaṭa ‘give’, the recipient alternates between a morphologically unmarked form that precedes the theme and a form marked with la- that follows it, in an alternation identical to the English double-object alternation. In that alternation, the la-marked form may not precede the theme, as (14) shows. The la-marked recipient of baʕat ‘send’, however, naturally precedes the theme, as (13) shows, suggesting that la- does not play the role of a preposition there, but rather that (13) exemplifies a double-object construction with a dative indirect object.

Cliticization facts parallel the contrast between (13) and (14). The la-marked recipient of baʕat in (13) pronominalizes in a morphological paradigm that includes the l of la-, as opposed to the accusative pronominal paradigm that the recipient of ʕaṭa instantiates. Pronominalization of the recipient argument in (13) is shown in (15a). As mentioned previously, the recipient argument of ʕaṭa never pronominalizes in the l-paradigm, as (12a) shows, repeated in (15b), even though it may in principle occur in a la-phrase, as (11a) shows. This again indicates that the la-phrase that ʕaṭa admits is not a dative DP but a PP that cannot be pronominalized.

The recipient argument of baʕat cannot be cliticized to the verb in the accusative paradigm and cannot be promoted to subject in the passive, as the examples in (16) show. This is typical of dative case, which is argued to be ‘lexical’ or ‘inherent’ in other languages (see e.g. Anagnostopoulou 2003; Andrews 1990; Czepluch 1996; Freidin and Sprouse 1991; Zaenen et al. 1985 on Germanic, Slavic and Greek). It is assigned in the base structure in a specific thematic configuration and therefore cannot be removed in the course of a derivation, for example by passivization.

The la-marked recipient of baʕat ‘send’ may follow the theme, as (17) shows, in which case it is unclear at the outset whether we are looking at a prepositional frame like that seen with ʕaṭa or inversion of the dative and accusative arguments of the double-object frame. It is clear that the theme of baʕat ‘send’ may cliticize to the verb in the accusative paradigm (17b) and promote to subject in the passive (17c), making it the ‘primary object’ by these criteria. Since we observed in the context of ʕaṭa ‘give’ that two object DPs may not be inverted, we assume that two object DPs of baʕat ‘send’ cannot be inverted either, and therefore that the examples in (17) exemplify the prepositional frame of the double-object construction.

Note lastly that although it is not possible to cliticize both objects of a double-object construction to the verb, it is still possible to pronominalize both objects: in this case, the first object cliticizes to the verb and the second is cliticized to a pleonastic place-holder stem jā- that the language employs for this purpose, as illustrated in (18a) and (18b) for ʕaṭa ‘give’ and baʕat ‘send’ respectively. The pleonastic stem jā- is very commonly employed in Damascus Arabic and some other dialects when for morphosyntactic reasons no other host is accessible.

The data above point to the following conclusion: a clitic pronoun must be affixed to a host, but affixation is subject to an intervention effect: no other potential target of cliticization may intervene between the clitic pronoun and the gap in the canonical position of the non-pronominal counterpart. The intervention effect means that in double-object constructions, a theme may only be cliticized to a verb in the prepositional frame, where it directly follows the verb, as (11b) shows. In the double-object frame, it is separated from the verb by the recipient and therefore not able to be cliticized to it, as (8b) illustrates. The theme may be pronominalized in that configuration if it is cliticized to the pleonastic stem jā. The role and etymology of jā is controversial. Bauer (1914) and Testen (1997/1998) trace it back to a proto-Semitic verb form, while Correll (1994) and Wilmsen (2013) trace it back to a proto-Semitic tonic pronoun (lacking the contemporaneous enclitic). Whatever its etymological source, it has undergone a high degree of grammaticalization. Correll points out explicitly that its contemporary distribution in Arabic is the same as that of a preposition, which would make jā+cl a PP, while Wilmsen characterizes the contemporary form jā+cl as a present-day tonic pronoun, somewhat similar to complex pronouns like ‘himself’, which would make it a DP. Whatever the proper decomposition of the form jā+cl is in modern Arabic, it is clear that nothing intervenes between the stem jā and the clitic cl, meaning the form obeys the condition on intervention in any case.

2.2 Clitic doubling in double-object constructions in Damascus Arabic

Clitic pronoun doubling in Damascus Arabic is slightly more restrictive than simple pronominalization. In transfer-of-possession constructions of the type discussed above in connection with pronominalization, only the recipient argument may be doubled. That is, a theme may not be clitic doubled in the presence of a clausemate recipient, even when the recipient appears in a prepositional phrase. The recipient blocks clitic pronoun doubling of the theme, regardless of whether it surfaces as a DP, as in (19a), or PP, as in (19b).^[8] This is in contrast to monotransitive constructions, in which doubling of the theme is natural, as illustrated in (2). In the examples discussed in this section, we underline the clitic and its full DP associate in order to clarify which object is being doubled.

Since cliticization of the theme over the recipient is not possible in the first place, the fact that clitic pronoun doubling is also blocked in this context, as (19a) shows, is not surprising. The surprising fact is that the theme can still not be doubled in the prepositional frame shown in (19b), where pronominalization of the theme (without doubling) is possible (compare (19b) with (11b)). The recipient does not act as an intervener for pronominalization in the prepositional frame, but yet seems to act as an intervener for clitic pronoun doubling in the same context. Surface adjacency of the theme to the host in the prepositional frame is not a sufficient condition for clitic pronoun doubling of the theme in that context, unlike pronominalization itself.

The recipient itself may be clitic doubled in ditransitive constructions. Since the preposition that marks the full DP double is the same preposition that marks the recipient in the prepositional frame, it is somewhat unclear at first glance which frame we are looking at in (20a). But since the clitic pronoun is accusative, which reflects the case of the recipient in the double-object frame (compare (20) with the basic cliticization pattern in (7b)), we infer that the recipient is being doubled as a DP here, not as a PP, and that the la- seen here is the la- that marks a full DP double of a clitic pronoun, not the preposition seen in Example (11). The fact that this la-marked recipient may also follow the theme, as (20b) shows, unlike the counterpart without la-, as (8a) shows, means that the la-marked double of the clitic pronoun has more flexibility in placement than the corresponding object in the non-doubling structure.

Circumstantial support for the conclusion that la-sāra in (20) is not a PP but a DP marked with the clitic doubling marker la- comes from Greek, where DP and PP recipient arguments are morphologically distinguished, and only DP arguments can be clitic doubled (Anagnostopoulou 2003; Dimitriadis 1999). DP recipient arguments bear genitive case and bear no special additional marking when clitic doubled. PP recipient arguments are introduced by the preposition se ‘to’. In this language, a genitive clitic pronoun can double a genitive DP recipient, as in (21a) (Anagnostopoulou 2003: 15, Example 18), but not a PP recipient, as (21b) shows (Anagnostopoulou 2003: 17, Example 22). If there is any crosslinguistic generality to the Greek facts, then (20), too, must represent doubling of an (in this case accusative) DP argument, even though the clitic pronoun doubling marker la- is homophonous with a preposition.

We mention in passing at this juncture that clitic pronoun doubling of recipient DPs is subject to the same definiteness restriction as clitic pronoun doubling of theme DPs illustrated in (3) and (4). Suñer (1988: 394–395) and Kallulli (2000: 212) report that in Spanish and Albanian respectively, clitic pronoun doubling of a recipient is not subject to a definiteness requirement. Suñer and Kallulli conclude that the clitic pronoun in those cases is an agreement inflection (obligatory in Albanian) rather than a full fledged pronoun. Damascus Arabic holds theme and recipient arguments to the same definiteness requirement in clitic pronoun doubling constructions. The semantic status of the clitic pronoun is uniform in all clitic doubling constructions in Damascus Arabic.

To return to the analytical problem that (19) poses, the fact that a theme cannot be doubled when a recipient is present stands in contrast not just to the theme in monotransitive constructions but also to theme arguments in locative constructions such as those illustrated in (22). Although locative constructions like those in (22) bear a syntactic resemblance to the prepositional frame of the double-object construction (they have the order V DP PP), they do not alternate with a double-object frame. We argue below that the two cases of V DP PP order have a different derivational source. In locative constructions, the theme may be clitic doubled, in spite of the clausemate PP. Apparently only PPs encoding recipient arguments block clitic pronoun doubling of the theme. Note that in Damascus Arabic, the word ward ‘flowers’ is a mass noun that triggers third person singular agreement. Note, too, that the prefix la- that occurs on the full DP double of a clitic pronoun may occur in the same sentence with the basic allative preposition la- ‘to’, as seen in (22b). This means that the constraint against doubling the theme in the presence of a recipient is not reducible to a superficial rule blocking two occurrences of la- in the verb phrase.

It is possible to clitic double the dative recipient argument of baʕat ‘send’, just as the recipient argument of ʕaṭa may be clitic doubled (Koutsoudas 1967), although as expected, the clitic pronoun is in the dative paradigm, reflecting the dative case of the recipient of baʕat ‘send’. Compare (23a) with (20a). The doubled recipient may follow the theme, as (23b) shows (Koutsoudas 1967: 512, Example 4). We conclude here as before that we are looking at the double-object frame in both cases, not the prepositional frame, since we concluded above that a recipient PP cannot be cliticized to the verb, much less clitic doubled. This conclusion parallel’s Suñer’s (1988) and Demonte’s (1995) conclusions for Spanish.

As with ʕaṭa ‘give’, the theme argument of baʕat ‘send’ cannot be clitic doubled when a recipient is present, as Koutsoudas also reports (see Koutsoudas 1967: 513, Example 28) regardless of order; compare (24) with (19). Again, this effect holds regardless of whether we construe la-sāra as a dative DP (as glossed in (24)) or a prepositional phrase.

The data above show that in Damascus Arabic, a theme may only be clitic doubled if no recipient is present at all, that is, in monotransitive constructions like (2) or in purely locative double complement constructions like those shown in (22), that designate movement of the theme along a path to a specified location, but not in change-of-possession constructions. When a recipient is present, though, in the form of an accusative or dative indirect object of ʕaṭa ‘give’ or baʕat ‘send’ respectively, this recipient blocks clitic pronoun doubling of the theme, as (19) and (24) show. In this case, only the recipient may be clitic doubled, as in (20) and (23). That pronominalization and clitic pronoun doubling are sensitive to the thematic role of the doubled DP is reinforced by the following point (though we will later reduce this thematic sensitivity to a structural generalization). Just as Rappaport Hovav and Levin (2008) show for English send, the counterpart baʕat in Damascus Arabic is ambiguous between a change-of-possession reading and a change-of-location reading. In its change-of-location reading, the preposition la- ‘to’ may be replaced by the preposition ʕala ‘on’ or ‘to’, commonly reduced to the prefix ʕa-. When baʕat takes a locative goal argument introduced by ʕala, it allows clitic pronoun doubling of the theme. The word katībe ‘division’ is feminine in (25), and is doubled by a feminine singular clitic pronoun. That is, in the context of one and the same verb (baʕat ‘send’), the theme may or may not be doubled depending on whether the other argument is construed as a location or a recipient.^[9]

The ungrammaticality of clitic doubling of the theme in change-of-possession examples like (19b) and (24b) is all the more puzzling because pronominalization is allowed in that same context. In the non-clitic doubled counterpart of (19b), shown in (11a) and repeated in (26a) the theme DP precedes the recipient PP and so is adjacent to the clitic host (the verb). Accordingly, the theme can be pronominalized as shown in (11b), repeated in (26b), since nothing intervenes between its canonical position and its host. Yet the theme cannot be clitic doubled here, as (19b) shows, repeated in (26c). And as described above, locative PPs do not block clitic pronoun doubling, even those built with the same preposition la- that PP recipient arguments occur with, as (22b) shows, repeated in (26d).

One possible – but we argue unfruitful – approach to the pattern in (26) is that clitic pronoun doubling is sensitive to the thematic context: when a recipient is present, only it may be doubled, independent of the intervention restriction on pronominalization. This would raise the question of what it is that makes recipient arguments special. One obvious thing that stands out about recipient arguments is that their instantiation as PP systematically alternates with an instantiation as DP preceding the theme. In this frame, they intervene between the theme and its potential host, the verb (see the schema in (9)). Another way of stating the restriction on doubling of a theme in the presence of a PP recipient, then, is that the prepositional frame acts for the purposes of intervention as if it were its alternant, the double-object frame, where the recipient intervenes between the theme and its host. This approach reduces the ungrammaticality of (19b)/(26c), repeated in (27a) below, to the ungrammaticality of (19a), repeated in (27b) below.

The fact that the prepositional frame behaves like the double-object frame in terms of intervention is the kind of constructional relatedness that in the structuralist tradition is taken to be a sign of transformational relatedness between the two formats. In fact, a body of literature on the double-object alternation claims that the prepositional frame is transformationally derived from the double-object frame (Aoun and Li 1989; Bowers 1981; Dryer 1986; Hallman 2015). Schematically, this derivation looks like (28a), where DP₁ names the recipient and DP₂ the theme. Locative constructions like those in (26d) and others previously mentioned are not derivatives of a double-object construction. They have the base order in (28b), where DP names the theme and PP the location.

The pattern in (27) lends support to this view. If the prepositional frame is derived from the double-object frame, the fact that the prepositional frame shows the same intervention effect for clitic pronoun doubling as the double-object frame can be captured as a level-ordering effect: unlike pronominalization, clitic pronoun doubling takes place before the derivation that converts the base double-object frame into the prepositional frame. At this point, the recipient still intervenes between the theme and its potential host, blocking doubling of the theme. Like pronominalization, then, clitic pronoun doubling is subject to an intervention effect. The difference between clitic pronoun doubling and pronominalization is that pronominalization may apply after the transformation in (28a), which puts the theme adjacent to the host.^[10] This analysis is purely structural and does not make reference to thematic roles. What is special about the recipient role in this analysis is that it is consistently base generated higher than the theme, and so intervenes between the host and the theme at the level of syntactic representation at which clitic pronoun doubling applies.

As a structural intervention effect, we do not expect the phenomenon described here to be sensitive to the particular thematic roles the objects of the verb bear, but only to the structural hierarchical order in which these arguments are found in the base structure. What matters for the analysis we propose here is not what thematic role the potential intervener bears, but where it is structurally situated in the base structure vis a vis the potential target of clitic pronoun doubling. As mentioned in Section 1, this expectation is borne out in another kind of double-object construction. We find the same intervention effect in a ditransitive causative construction in Arabic in which causee arguments function as interveners for theme arguments. Many transitive verbs, such as libis ‘put on (an article of clothing)’, ʕirif ‘know about’, akal ‘eat’ and others have causative counterparts that share the syntactic behavior of change-of-possession verbs described above. In their monotransitive base form, these verbs, as expected, allow clitic pronoun doubling of a theme, as (29) illustrates.

These verbs can be causativized, adding a causer to the set of arguments in the transitive base. Causativization is typically morphologically marked templatically, by doubling the middle consonant of the (usually) three-consonant root of the underlying verb, though the causative counterpart of akal ‘eat’ is the distinct lexeme ṭaʕma ‘feed’. The data in (30) below show that the internal arguments of the causative counterparts of the verbs in (29) may surface either in a double-object frame seen in the (i)-examples, or a prepositional frame seen in the (ii)-examples, exactly like we find with change-of-possession verbs like ʕaṭa ‘give’.

These causative verbs allow clitic pronoun doubling of the causee argument, as seen in the (i)-examples below, but not the theme, regardless of whether the causee occurs as a direct argument of the verb in the double-object frame, as seen in the (ii)-examples below (compare with the (i)-examples in (30)) or in the prepositional frame, as seen in the (iii)-examples below (compare with the (ii)-examples in (30)).

This is the same grammaticality pattern we see with change-of-possession verbs, where the theme cannot be clitic doubled in the presence of a recipient, even when the recipient is in a prepositional phrase, even though in that case the theme is surface-adjacent to its potential host, the verb. The only difference is that in (31) what blocks clitic doubling of the theme is not a recipient but a causee, again regardless of whether the causee is a direct argument or occurs in a prepositional phrase. But in the case of causative constructions, it is clear that the underlying order is one in which the causee is structurally superior to the theme, since the causee is the agent of the underlying transitive construction that is causativized in (29)–(31). Given a basic transitive construction like (32a) where DP₂ names an agent and DP₃ a theme, causativization can be schematically represented as in (32b), where DP₁ names a causer of an event described by the configuration in (32a), whose thematic structure remains intact in (32b). Raising of the main verb to the causativizing morpheme cause derives the causative morphological alternant of the basic monotransitive verb as well as a syntactic structure that feeds the double-object alternation rule in (28a), converting the (i)-examples in (30) into the (ii)-examples.

The analysis proposed here that clitic pronoun doubling is sensitive to the base structure of the ditransitive construction, where recipient and causee arguments are syntactically superior to theme arguments and therefore closer to the host, captures the parallel behavior of causative and change-of-possession constructions. The fact that causative constructions show the same set of transformations and restrictions that change-of-possession constructions do corroborates the idea that clitic pronoun doubling is not sensitive to the thematic environment per se, but rather to the syntactic hierarchy: recipient and causee arguments are higher in the structure and therefore closer to the morphological host than theme arguments, which in turn are higher than location arguments. The intervention restriction on pronominalization dictates then that recipient and causee arguments interfere with clitic pronoun doubling of a theme, even in the form of a PP, while location PPs do not. This analysis is summarized in (33).

The fact that this analysis presents an explanation for the puzzling intervention of a prepositional recipient in clitic pronoun doubling of a theme in spite of the surface adjacency between theme and host conversely supports the analysis of the double-object alternation illustrated in (28a), where the double-object frame serves as base structure for the prepositional frame. Other analyses of this alternation are not able to capture the intervention of a PP recipient for clitic doubling of a theme, at least not in purely structural terms, as we have done here. The early analyses of the double-object alternation by Perlmutter and Postal (1984) and Larson (1988, 2014 claim that the derivation goes in the opposite direction: the prepositional frame is the base from which the double-object frame is derived (by inversion of the objects of the verb and deletion of the preposition). In this analysis, the theme is hierarchically higher than the recipient in the base structure, which surfaces with no change in the prepositional frame. This structure makes the prepositional frame identical in structure to locative constructions, offering no explanation for why recipient PPs behave differently from locative PPs in interrupting clitic pronoun doubling of a theme in Arabic.

A similar problem faces a more recent view of the double-object alternation, which claims that the two frames are not transformationally related. Harley (1995, 2002, Bruening (2001, 2010 and many others claim that the double-object frame has the base order recipient > theme (both DPs) while the prepositional frame has the base order theme > recipient (the recipient a PP). The fact that here, too, the prepositional frame is base generated as such attributes to this frame the same structure as locative constructions and leaves us with no account of the difference between recipient/causee PPs and location PPs in whether they block clitic pronoun doubling of a theme. The analysis proposed here, that recipient/causee arguments are base generated higher than theme arguments, even when they ultimately appear in a prepositional phrase, makes it possible to capture their inhibitory effect on clitic pronoun doubling of a theme as a rule ordering effect. This analysis captures the uniform behavior of change-of-possession and causative constructions under one generalization, and captures the distinction between pronominalization and clitic pronoun doubling in a parametric manner: both are subject to an intervention restriction, but clitic pronoun doubling is ordered before the derivation of the prepositional frame, while pronominalization is not. With these conclusions about Damascus Arabic in mind, we turn to the pronominalization and clitic pronoun doubling facts in Al-Bāṭinah Arabic.

3.1 Pronominalization in double-object constructions in Al-Bāṭinah Arabic

Like Damascus Arabic, Al-Bāṭinah Arabic has two classes of double-object verb, that take an accusative and dative recipient respectively. We begin as before with the double-accusative class of verbs, exemplified in Al-Bāṭinah as in Damascus Arabic by ʕaṭa ‘give’. As in Damascus Arabic, the double-object construction in (42a) alternates with a prepositional construction where the recipient is marked by the preposition l-, seen in (42b). This preposition is cognate with Damascene la- but lacks the vocalic rhyme, surfacing as a syllabic [l] instead.^[12]

Unlike Damascus Arabic, the objects can be freely reversed in the double-object frame. While the order recipient > theme seen in (42a) is regarded as canonical, the other order, illustrated in (43), is acceptable as well.

This optionality goes hand in hand with an optionality in which object may raise to subject in the passive. Just as in Damascus Arabic, the recipient may cliticize to the verb and raise to subject in the passive, as illustrated in (44). But unlike Damascus Arabic, so may the theme, as illustrated in (45).

In the prepositional frame, the theme may cliticize to the verb and may raise to subject in the passive, as (46) illustrates. This is as in Damascus Arabic.

Like Damascus Arabic, Al-Bāṭinah Arabic also has a class of double-object verbs that take a dative indirect object, also marked by l-. The verb rasal ‘send’ is characteristic of this class. Example (47a) translates Damascus Arabic (13) into Al-Bāṭinah Arabic. The dative recipient can be cliticized to the verb as a pronoun in a dative inflectional paradigm, illustrated in (47b).

The dative recipient cannot be cliticized in the accusative paradigm nor promoted to subject in the passive, as illustrated in (48). This is as in Damascus Arabic (16) and is, as mentioned in Section 2.1, typical of dative arguments crosslinguistically.

As in Damascus Arabic, the accusative and l-marked internal arguments in (47a) can in principle be reversed, as shown in (49a), but because of the ambiguity of l- between a preposition and a marker of dative case, it is unclear whether we are looking at a prepositional frame here or an inversion of the dative recipient with the accusative theme. Given the availability of this inversion with two accusative objects in Al-Bāṭinah Arabic, there is no reason to think that this is not possible when one of the objects is dative, so we assume (49a) is ambiguous between a prepositional frame and an inverted double-object frame with a dative recipient. Nothing hangs on this conclusion, though. It is also difficult to resolve the question of whether the dative recipient blocks cliticization of the accusative theme. Cliticization of the theme and raising to subject in passive are possible with rasal, as (49b) and (49c) make clear, but it remains unclear whether the l-marked recipient in those examples is a PP or a dative DP that is a non-intervener for cliticization of the theme to the verb.

Al-Bāṭinah Arabic uses the same strategy as Damascus Arabic to express two pronominal objects in the double-object frame, namely insertion of jjā- to host the object clitic pronoun, shown in (50).

Al-Bāṭinah Arabic appears to be a great deal more liberal than Damascus Arabic in terms of both word order within the verb phrase and the possibilities for pronominalization. Any DP object of the verb may be cliticized to it (but only one at a time, as in Damascus Arabic). Damascus Arabic lacks this flexibility. We return to this matter in Section 4, where we suggest that Al-Bāṭinah Arabic lacks the intervention restriction that is operative in Damascus Arabic, but turn first to clitic pronoun doubling in double-object constructions in Al-Bāṭinah Arabic.

3.2 Clitic doubling in double-object constructions in Al-Bāṭinah Arabic

In the double-object frame in Al-Bāṭinah Arabic, either argument may be clitic doubled, regardless of order. This is unlike Damascus Arabic but of course correlates with the fact that in Al-Bāṭinah Arabic, either object can precede the other and accordingly either may be cliticized to the verb or raised to subject in the passive. The examples in (51) show clitic pronoun doubling of the recipient in both orders, while (52) shows clitic pronoun doubling of the theme in both orders. Recall that in Damascus Arabic, a theme may not be doubled when a recipient is present. The examples in (52) show that this restriction does not hold in Al-Bāṭinah Arabic. We note here, as promised in Section 3, that the clitic doubled DP in (51a) and (52b) is not clause-final, in contrast to the example of clitic right dislocation in French in (39). Clitic doubling in Al-Bāṭinah Arabic does not involve right dislocation of the DP associate of the clitic pronoun, and is therefore not reducible to clitic right dislocation. As in our discussion of Damascus Arabic, we underline the two terms that stand in the doubling relation in the Al-Bāṭinah examples below.

The theme may also be doubled in the prepositional frame. This is again unlike Damascus Arabic, where the mere presence of a recipient in the verb phrase blocks clitic pronoun doubling of a theme, regardless of which frame we are looking at. Compare (53) with Damascus Arabic (19b).

As in Damascus Arabic, a theme may also be clitic doubled in the presence of a locative PP designating the end of a path along which the theme moves.

When we turn to the class of dative-assigning verbs exemplified by rasal ‘send’, we find that the recipient may be clitic doubled, but not as a dative (i.e. l-marked) DP, but as a bare DP, as is typical for all the clitic pronoun doubling contexts we have seen in Al-Bāṭinah Arabic so far. That is, the morphological dative case of the clitic pronoun is not reflected on the full DP double. The full DP double is always bare regardless of the morphological case of the associated clitic pronoun. The full DP double of the recipient clitic pronoun may precede or follow the theme.

Here again, it is possible to clitic double a theme in the context of a l-marked recipient of rasal, as shown in (56a), again unlike Damascus Arabic. Here, too, the order of the double with the other argument is immaterial, as (56b) shows. In these respects, rasal is parallel to the way ʕaṭa ‘give’ behaves in Al-Bāṭinah Arabic (compare (56a) with (52)).

In light of the parallels between change-of-possession constructions and causative constructions we have observed in Damascus Arabic, we expect the same flexibility of order, pronominalization and clitic pronoun doubling seen in change-of-possession constructions in Al-Bāṭinah Arabic to carry over to causative constructions. Al-Bāṭinah Arabic does indeed show the same causativization paradigm as Damascus Arabic and the same syntactic flexibility as it displays in change-of-possession constructions. Since these facts are as expected, and since they do not point to any restrictions crucial for the analysis of pronominalization and clitic pronoun doubling in Al-Bāṭinah Arabic, we do not address them here, but list them for the record in Appendix B. In summary, there is no apparent difference between the circumstances under which a DP may be doubled by a clitic pronoun and the circumstances under which it may be pronominalized in the first place in Al-Bāṭinah Arabic.

Appendix A: Attested clitic pronoun doubling examples in Al-Bāṭinah Arabic

The following is a selection of examples of clitic pronoun doubling in Al-Bāṭinah Arabic attested in Youtube videos. However, videos for all but the first five examples are no longer available at the time of publication. When available, the videos can be viewed at the URL given under each example. The example occurs in the time frame given to the right of the URL. Examples (65)–(67) are all from the same clip.

Appendix B: The Al-Bāṭinah Arabic causative pattern

The following examples show word order and clitic pronoun doubling patterns in transitive and causative alternants of the verbs labas ‘put on’, ʕaraf ‘know, find out’, and ʔakal ‘eat’ in Al-Bāṭinah Arabic, parallel to the facts reported for Damascus Arabic in the discussion surrounding Examples (29)–(31). All the sentences listed below are grammatical in Al-Bāṭinah Arabic.