The literature on resumptive pronouns has argued that they can be of two types: (a) those that appear in base-generation dependencies, and (b) those that terminate movement chains (Scott 2021; Georgi & Amaechi 2023; Hewett 2023; Yip & Ahenkorah 2023; Ershova 2024; Fongang 2025, i.a.,). RPs of the (b)-type, it has also been demonstrated, tend to differ from those of the (a)-type in that they do not always match or fully match moved XPs in phi-features.1

In Swahili (Bantu), for example, copies of moved pronouns do not always match them in person features. The regular first-person singular RP mi, in the parasitic gap construction in (1), is ungrammatical with the moved first-person singular pronoun Mimi ‘me’. Only ye which, as shown in Scott (2021), is unmarked for person features is allowed.

Georgi & Amaechi (2023) also argue extensively that in Igbo (Niger-Congo, Nigeria), moved pronouns, in the focus construction in (2) for example, can only be resumed by third-person singular yá, regardless of person and number values. They, as a consequence, demonstrate that Igbo movement RPs show a number and person mismatch.

Some of the types of mismatches that have been identified within a single language in the recent literature are (a) person mismatches (e.g., Swahili and Samoan; Scott 2021; Ershova 2024), (b) person and number (e.g., Igbo and Cantonese; Georgi & Amaechi 2023; Yip & Ahenkorah 2023), and (c) number and gender (e.g., Akan; Yip & Ahenkorah 2023).

In this squib, I present novel data from Cameroon Pidgin English (CPE) to show that movement RPs can also show a person-gender-number mismatch.2 I demonstrate that CPE has movement and base-generation RPs. Unlike base-generation RPs, movement RPs show a number-gender mismatch with lexical object XP foci, and a number-gender-person mismatch with pronominals. I relate this to the recent literature and argue that, empirically, number-gender-person mismatches are attested within a single language and, theoretically, the pattern in CPE is predicted by copy-deletion (Van Urk 2018; Scott 2021; Georgi & Amaechi 2023; Yip & Ahenkorah 2023) and impoverishment (Ershova 2024; Fongang 2025) approaches to movement RPs.

The squib is structured as follows: §2 presents the CPE data in detail. §3 demonstrates that CPE has movement as well as base-generation RPs. In §4, I discuss the consequences of the contents of §3 for recent theories of φ-mismatches in resumption, and show how such approaches can derive the CPE data. §5 concludes.

In CPE, resumptive pronouns are selected from the paradigm of personal pronouns in Table 1 below (see also Atindogbé & Chibaka 2012 and Ayafor & Green 2017).3

The examples in (3) and (4) illustrate the use of 3rd person pronouns in cross-sentential anaphora contexts. In (3-a), for example, the inanimate XP moto ‘car’ cannot be replaced by the 3rd-person-singular-animate pronoun yi. Only am is grammatical.4 When the nominal antecedent is animate (c.f., (4-a)), the grammatical pronoun is yi, and not am. Both inanimate (3-b) and animate (4-b) plural nouns need to be substituted by dem, and it shows up in subject and object positions.5

What is puzzling is that in object XP focus, the only grammatical resumptive pronoun is am.6 This is so with animate-singular (5-a), animate-plural (5-b), inanimate-singular (6-a) and inanimate-plural (6-b) nouns.

The focus-marked object XPs in (5) and (6) are lexical. Pronoun focus also requires am in the resumption site, irrespective of the person, gender and number values of the focused pronoun (7).7

Interestingly, am also shows up in object topicalization, but only if the topic XP is singular-inanimate (8), as one would expect.

Topicalization of animate XPs in object position triggers resumption by corresponding, matching, pronouns (9).

Pronominal object XP topics are also resumed by corresponding matching pronouns. I illustrate this in (10-a) with singular pronouns, and in (10-b) with plurals.

Unlike the other pronouns in Table 1, am can neither be focused (11) nor topicalized (12); a property of weak pronouns (Manzini 2014, i.a.,).

The topicalization and regular-pronoun-use data suggest that am is a 3sg inanimate object pronoun, albeit possibly weak (c.f., (11) and (12)). The fact that it is the only grammatical RP in object XP focus contexts, irrespective of number and animacy (c.f., (5) and (6)), demonstrates that CPE has a RP that shows a number-gender mismatch. Besides, the language instantiates a person-number-gender mismatch with focused pronouns in object position (c.f., (7)). With this in mind, this squib, empirically, adds to the discussion the fact that resumptive pronouns can show a person-number-gender mismatch in a single language.

The recent literature on resumption has argued that while movement RPs are created by the dependency itself, base-generation RPs result from binding. Scott (2021); Georgi & Amaechi (2023) and Yip & Ahenkorah (2023), building on Van Urk (2018), argue for a copy-deletion approach to movement RPs. In such approaches, the head (the moved XP) and its copies are identical. Economy (Landau 2006) and P-Recoverability interact to force deletion of parts (or all) of the structure of the copies, yielding resumptive pronouns (or gaps). The locus of features and the amount of structure that gets deleted can, as a consequence, create non-matching or partially-matching RPs.

Ershova (2024), building on Baier (2018), proposes an approach to non-matching movement RPs that relies on Impoverishment. Specifically, she shows that non-matching RPs in Samoan resumptives result from impoverishment rules that are triggered by the simultaneous presence of φ and Ā-features on adjacent heads. In the remainder of this squib, I show that the CPE data are predicted by these two approaches. Before laying out the details of each proposal, it is important to demonstrate that object XP focus, which requires the non-matching RP am, involves movement and not base-generation. I do this in the next section.

This section provides empirical evidence that CPE has two types of RPs: movement and base-generation. I demonstrate that object XP focus, unlike topicalization, involves movement in CPE. For space limitations, I will only present two diagnostics, namely island-sensitivity and idiom chunk reconstruction.8

Section 3 has demonstrated that CPE object XP foci move into the focus position. What this means for copy-deletion approaches to pronoun copying and resumption (Van Urk 2018; Scott 2021; Georgi & Amaechi 2023; Yip & Ahenkorah 2023) is that the relevant feature mismatch with am, as described in Section 2, can be thought of as involving a structure-reduction algorithm. The locus of features and the amount of structure that gets deleted can, as a consequence, create the non-matching RP. One major difference between these approaches is what counts as a deletion domain. In Van Urk (2018), deletion targets phases, and nP and KP are phase boundaries (in, at least, Dinka Bor). In Scott (2021) and Georgi & Amaechi (2023), the deletion domain is dynamic, and is conditioned by MaxElide (c.f., Landau 2006) which, in turn, obeys language-specific requirements.

Van Urk (2018), for example, shows that in Dinka Bor, when the XP tó̤o̤ny ké díi ‘how many pots’ in (19) undergoes wh-movement, it leaves full copies in every intermediate movement site. These copies are then subject to a deletion algorithm that produces all the instances of ké in (19).

What is interesting about Dinka Bor, he demonstrates, is that a second-person plural topic pronoun will also be copied by ké, as illustrated in (20).

Example (20) also shows that there is a person mismatch between the topic pronoun and its copy. A second-person topic pronoun is copied by a third-person pronoun. Van Urk (2018) argues that ké is unmarked for person features, and its structure is as illustrated in (21). Person features are located on the n head, and the structure of the second-person-plural pronoun wêek is given in (22).

When wêek undergoes topicalization in (20), it leaves a full copy in the movement site. The moved pronoun and its copy have the structural representation in (22). Economy (23) and P-Recoverability (24) then interact to force partial deletion of parts of the structure of the lower copy, i.e., the copy in the movement site.

Van Urk (2018) argues that phasehood determines the amount of structure that can be deleted. In Dinka Bor, the relevant phase-defining heads are n and K. In deriving the presence of ké in (20), he proposes that nP gets deleted. Since its head n hosts person features, the resulting structure corresponds to that of ké. As a result, the second person pronoun topic wêek is copied by ké.

There also are approaches to φ-mismatches in resumption that do not rely on copy-deletion. One such is proposed by Ershova (2024). She argues, based on data from Samoan, for an approach to φ-mismatches in movement RPs that relies on Impoverishment. Specifically, building on, among others Baier (2018), she proposes that Ā-features can interact with φ-features and, as a result, trigger impoverishment. The consequence of this is that non-matching or partially matching RPs can be inserted in the resumption site. Both types of approaches assume that the lower copy of a fronted lexical DP is pronounced as a pronoun due to the absence of the nominal root, as schematized in (25) and (26).

While a structure-deletion approach deletes nP to account for the pronoun copy example in (20), the impoverishment approach would simply remove the person feature on the n head, in the presence of an Ā-feature.

The two approaches that precede, it appears, can be used to account for am-resumption in CPE. As a quick reminder from Section 2, am is the only grammatical RP in object XP focus, irrespective of the person, gender and number values of the focused XP. To show how the algorithms would proceed, I first propose a featural decomposition of the relevant CPE personal pronouns, to the exclusion of quirky am. The list is given in (27) below. (27) is based on Table 1, repeated here as Table 2.

Underspecification and the Subset Principle (28) derive the syncretic exponents in the paradigm of CPE pronouns. Second person yu, for example, is syncretic in the nominative and accusative because it is underspecified for case features. Third person nominative i can resume animate as well as inanimate singular subjects because it is underspecified for gender. To account for the fact that am is the only possible RP in object XP focus (as shown again in (29) and (30)), it seems empirically logical to propose that is is only specified for case (31). The relevant case feature is accusative, because am never resumes subject XPs (c.f., Section 2).

Under both copy-deletion and impoverishment approaches, the focused pronoun mi ‘me’ (30), for example, leaves an identical copy in the movement site. The structural representation of mi is given in (32) and that of am in (33).

In line with the copy-deletion approach, NumP in (32) will be deleted, to ensure that the corresponding structure is that of am.10 For the impoverishment-based account, one simply has to assume that the impoverishment rule in (34) deletes person and number features in the presence of the focus feature, and in object contexts.

Overall, the pattern in CPE is predicted by both approaches to φ-mismatches in resumption.

This squib investigated the quirks of resumption in CPE and showed that the language instantiates a person-number-gender mismatch. It also demonstrated that CPE is similar to, for example, Akan, Cantonese, Igbo, Samoan and Swahili in that it has both movement and base-generation RPs. Only movement RPs, the paper argued, allow feature mismatches between the moved item and the RP. Overall, the squib shows that both impoverishment and structural deletion predict that we should find a language that reduces/deletes all φ-features in a particular context. CPE instantiates this pattern. While this does not help distinguish between the two theories, it contributes to the empirical landscape in an important way.