Low topics are not IP-external: a reply to Alzayid (2025)

1 Introduction

Alzayid (2025) contends that, while Jarrah and Abusalim’s (2021) analysis of the low IP area in Jordanian Arabic (JA) and Alshamari and Jarrah’s (2022) account of low topics in Najdi Arabic (NA) are “attractive” (Alzayid 2025: 1415), “there are somber prospects for the low IP analysis when applied to right-dislocation in NA and JA, and that an IP-external analysis is warranted” (Alzayid 2025: 1408). Essentially, Alzayid (2025) presents three arguments against Alshamari and Jarrah’s (2022) analysis of right-dislocated elements in NA as low elements (see Belletti 2004), advocating an IP-external configuration for such elements. However, as we demonstrate below, the three arguments advanced by Alzayid (2025) in support of the IP-external analysis of right-dislocated elements in NA are either irrelevant or not compelling. Additionally, it should be noted that the purported failure of the cartographic analysis for the right-dislocated elements in NA as implicitly suggested by Alzayid (2025) does not, ipso facto, substantiate the IP-external hypothesis for these elements. In this reply, we discuss the arguments provided by Alzayid (2025) that claim the invalidity of Alshamari and Jarrah’s (2022) account of right-dislocated elements in NA as low topics, dismissing them based on independent evidence.

2 Evaluating the NPI argument

The first argument supplied by Alzayid (2025) – that what Alshamari and Jarrah (2022) treat as low topics should be rather IP-external – is based on the behaviour of N(egative) P(olarity) I(tems). Alzayid (2025) underscores the requirement that NPIs must appear within the scope of neg-words for proper licensing. For instance, in the following two examples the negation marker ma ‘not’ is obligatory for the licensing of the NPIs walaw ‘even’ and ʕmur ‘never’, respectively.

Alzayid argues that the fact that NPIs walaw ‘even’ and ʕumur ‘never’ cannot appear right-dislocated in the following two examples is evidence that NPIs are not located low in the tree; otherwise, the c-command condition for their licensing would be satisfied.

For Alzayid (2025), NPIs walaw ‘even’ and ʕmur ‘never’ should be located in an IP-external position, in which case the c-command condition of these two NPIs is not met. Therefore, what appears as low topics would be located in a structural position which does not fall within the c-command domain of higher elements such as neg ⁰.

However, the ungrammaticality of the examples in (2) arises from an independent factor that does not conclusively determine whether right-dislocated elements function as low topics or IP-external constituents. This factor is that NPIs are indefinite/nonspecific elements. To clarify this point, we must first address the issue that Alzayid (2025) incorrectly treats the clitic on the verb in the two examples below as obligatory, which contradicts the original source from which these examples are drawn (Alshamari and Jarrah 2022):

Alshamari and Jarrah (2022: 1016) report that “NA exhibits highly productive instances of a construction where the clitic still occurs on the verb while the object appears postverbally”, as shown in the following example:

Alshamari and Jarrah (2022) assert that in contexts where the clitic of the postverbal object appears on the verb, the object must be definite or specific. Alshamari and Jarrah (2022: 1016) mention that “[o]ne piece of evidence that indicates an intimate link between the presence of this object doubling clitic on the verb and the givenness of the object comes from the incompatibility of this clitic with nonspecific, indefinite objects”. When the object is an indefinite, nonspecific element, the clitic on the verb is prohibited, as evidenced by the following two examples (Alshamari and Jarrah 2022: 1017):

The correlation between the presence of the clitic of the postverbal object on the verb and the definiteness/specificity of the object is considered evidence that the object in these cases functions as a topic. A significant point to highlight here is that NPIs are generally treated as negative indefinites. In this regard, Alsarayreh (2012: 6) mentions that “[t]he term negative indefinites is used to capture the idea that these items [NPIs] can contribute both indefinite and negative meaning to the utterance in which they occur”. Accordingly, the fact that NPIs cannot be associated with a clitic on the verb while they appear postverbally is not due to the failure of the c-command condition for their licensing, but because they are indefinite elements.

Notice here that the second example presented by Alzayid, shown in (2b) above and reproduced below for convenience, which is meant to support his argument that NPIs cannot be dislocated, is ungrammatical because the NPI ʕmur is right-dislocated without an associate.

Alsarayreh (2012: 62) mentions that ʕumur is a “temporal indefinite adverb”, which therefore cannot be resumed on the verb yru:h. In Arabic, only nominals can be resumed on heads.

Given this, NPIs cannot serve as reliable diagnostics for the syntactic position of right-dislocated elements. Some NPIs are adverbs, which inherently resist verb resumption, while others are indefinite constituents, making them incompatible with the resumption conditions placed on postverbal objects, which require definiteness or specificity.

3 Agreement and VOS word order

The second argument presented by Alzayid (2025) to support his claim that there is no space for low topics in the core IP structure is based on the observation that in VOS constructions, the verb shows full agreement with the post-object subject, both in JA (7a,b) and NA (8a,b), as demonstrated in the following examples (Alzayid 2025: 1412–1413):

Alzayid argues that, given the position of the subject to the right of the object, the subject cannot be located in Spec,TP (or SpecIP, in his terms). To account for the full agreement between the subject and the verb while the subject is right-dislocated, Alzayid proposes three scenarios. The first scenario suggests that the subject goes through SpecIP to check agreement features. The second scenario posits that the rich agreement in (7a) and (8a) occurs because a preverbal pronominal pro doubles the dislocated subject and simultaneously establishes “an agreement relation with I⁰” (Alzayid 2025: 1413). The third scenario, described as “straightforward” (Alzayid 2025: 1413), simply assumes “an adjunct status for the subject […] implying that there is no room for this element in the skeleton of the core IP” (Alzayid 2025: 1413).

Alzayid ends his discussion with an open question: “Can the low IP analysis be upheld under such a scenario?” Our response is that the low IP analysis can certainly be upheld. In fact, there is no issue with this. The derivation of VOS constructions (in JA), where full agreement between the subject and the verb is maintained, is discussed by Jarrah and Abusalim (2021). They argue that in such cases, the subject is a low focus that merges in Spec,vP and then moves to Spec, FocusP in the low IP area. Jarrah and Abusalim (2021: 146) provide the following tree as a possible derivation of VOS constructions (note that the verb would be adjoined to T0, which has unvalued φ-content, but T0 is deleted from the tree for convenience):

(9)

In such constructions, T⁰, where the verb is adjoined, establishes an Agree relation with the subject in a probe-goal configuration. According to the Agree operation, the probe, which is in this case T⁰, and the goal, i.e., the subject, do not need to be in a Spec-head relation for the Agree relation to occur. For Chomsky (2001), the probe c-commands the goal, allowing the goal to value the probe’s unvalued φ-content, as shown in the following tree:

(10)

Therefore, full subject-verb agreement in VOS constructions, where the verb is assumed to be adjoined to T⁰, does not undermine the low IP topic analysis. Instead, it serves as a strong counterargument against the external-IP configuration. In this configuration, positing a probe-goal relation between T⁰ and the IP-external subject creates theoretical difficulties unless we assume the presence of a non-expletive referential pro in Spec,vP, with which T⁰ enters an Agree relation. However, this assumption leads to complications by over-generating instances of null elements without sufficient empirical evidence to support them.

4 Wh-phrases and focus structure

The final piece of evidence presented as supporting an external-IP analysis is based on the interaction between wh-phrases and foci in JA and NA. Alzayid (2025: 1414) provides two examples where a focus element and a wh-expression occur together:

Alzayid mentions that the right-dislocated elements in the two examples above, i.e., the wh-phrase mi:n ‘who’ in JA and min ‘who’ in NA, should be generated IP externally given that the wh-phrase must not be embedded under another focus, which is in the two examples the object of the prepositions AMMAN and RIYADH, respectively. For Alzayid (2025), the wh-phrases are right-dislocated. However, the two examples provide more insight into this issue. If we submit that the wh-phrases are right-dislocated, the rest of the examples would be right-dislocated as well as they appear to the right of the wh-word. The embedded clause is therefore entirely right-dislocated, except for the focused phrase which appears clause initially. A more cautious assumption would be to propose that the contrastively focused element the PP fi AMMAN ‘in Amman’ or bi RIYADH ‘in Riyadh’ move to the left periphery of the embedded clause, while no right-dislocation takes place. This assumption is supported by the fact that contrastive focus, unlike information focus, is morphologically marked by overt movement in Arabic grammar (Moutaouakil 1989, among others). Additionally, the constructions mentioned in the examples in (11) are not comparable to those discussed by Alshamari and Jarrah (2022), where elements that are neither interrogative nor contrastive in nature are examined. It is theoretically untenable to extend the analysis of wh-words to entirely distinct categories, such as topics, whose distribution and syntactic behaviour may diverge significantly from those associated with focus or interrogative elements.

5 Conclusions

In conclusion, Alzayid’s (2025) three arguments attempting to substantiate the claim that low topics are right-dislocated and, hence, IP-external fail to withstand scrutiny. Each argument proves either irrelevant or not compelling enough to dismiss the assumption that low topics are IP-internal. Indeed, these arguments collectively reinforce the original proposal, i.e., low topics are structurally positioned between TP and vP, thereby validating the presence of a low discourse-related area between TP and vP.