1. Introduction
Alongside regular second-person imperatives like Gaan huis toe! ‘Go home!’, where the addressee and the performer of the action are one and the same, both Afrikaans and Dutch also allow imperatives in which the addressee and the performer are two separate entities. Consider the permissive let-imperatives in (1) and (2).
In these examples, the command is directed at the addressee, while the described action is carried out by someone else, as indicated by the presence of a first- or third-person oblique-marked pronoun. Despite involving a different performer, these let-imperatives still function as directives aimed at the addressee, much like regular imperatives.
It is, however, quite common across languages to find imperative-like constructions that do not have the addressee-oriented, directive meaning that we expect from imperatives (Holvoet Reference Holvoet2020). Both Afrikaans and Dutch feature V1-constructions with the verb laat/laten ‘let’ that cannot be classified as commands, but that have the same form as some of the let-imperatives that we find in these languages. Consider the Afrikaans examples in (3) and the Dutch examples in (4).
Instead of giving a command that needs to be complied with as in (1) and (2), the examples in (3) and (4) serve to indicate the speaker’s surprise, amazement, or even shock about a certain event or action. From now on, I will refer to non-commanding let-constructions like the ones above as pseudo-let imperatives or pli-constructions for short.
The pseudo-let imperatives in Afrikaans and Dutch express two kinds of meanings: a hypothetical one and a mirativeFootnote 1 one. The hypothetical meaning arises when the speaker wants to express a fact that would hold in a hypothetical situation, as in (3a) for Afrikaans and (4a) for Dutch. These PLIs are similar in interpretation to the conditional uses of imperatives that we see across languages (Aikhenvald Reference Aikhenvald2010:237). The conditional imperative is used by the speaker to get the addressee to imagine a specific situation, and to present the inevitable outcomes or result of that situation (Holvoet Reference Holvoet2020:336). For example, in the conditional imperative Leave out the flour and your cake will be a disaster!, the speaker sketches the condition or the hypothetical situation (leaving out the flour), and follows it with the expected consequence of that specific situation (the cake being a disaster). PLI-constructions with a hypothetical meaning behave similarly in this respect. In (3a), for example, given a hypothetical situation where the speaker enters a shoe store, the resulting action, the buying of more shoes, is presented as a given. A similar idea holds for the Dutch example in (4a).
The mirative PLI, on the other hand, is used to convey the speaker’s surprise or amazement about a certain event or action that has actually taken place, as in (3b) for Afrikaans and (4b) for Dutch. As such, unlike the hypothetical PLI, mirative PLIs do not refer to a hypothetical situation, but rather to real-world events – specifically, ones that are unexpectedly true. For instance, (3b) conveys the speaker’s disbelief or shock about the fact that she has just dropped the cake that was meant for a party starting in 10 minutes. Similarly, in the Dutch example in (4b), the speaker is surprised or amazed by the unexpected coincidence that she has finished her book just as the possibility of a raise comes into play.
The goal of this article is twofold: Firstly, I show that the PLI-laat/laten ‘let’ has undergone grammaticalization in both Afrikaans and Dutch by drawing on criteria used by Van Craenenbroeck & Van Koppen (Reference Craenenbroeck and van Koppen2015, Reference Craenenbroeck and van Koppen2017) in their work on the grammaticalization of perception and causative verbs in Dutch. Furthermore, I show that Dutch PLIs do not have all the same options as their Afrikaans counterparts, and thus, that Afrikaans PLI-laat has grammaticalized further than its Dutch counterpart. Secondly, I explore whether this grammaticalization is likely due to purely internal changes, or whether language contact may also have played a role here. I will suggest a possible explanation for why Afrikaans PLI-laat has grammaticalized further than its Dutch counterpart by proposing that grammaticalization as an internal change played a vital role in the development of the PLI-constructions in both languages, but that Afrikaans’ contact with other related and unrelated languages throughout its history may have fed into its further grammaticalization.
This article is structured as follows. In section 2, I discuss key characteristics of the PLI-constructions introduced above, harnessing the criteria used by Van Craenenbroeck & Van Koppen (Reference Craenenbroeck and van Koppen2015, Reference Craenenbroeck and van Koppen2017; henceforth VCVK) to examine the grammaticalization of PLI-laat/laten in Afrikaans and Dutch. Thereafter, in section 3, I explore the question of whether these changes occurred because of internal linguistic factors, contact with other languages, or a combination of the two. Section 4 concludes.Footnote 2
2. Grammaticalization of PLI-laat/laten
The process of grammaticalization canonically involves lexical items becoming more functional through regular use in specific grammatical contexts (see, among others, Hopper & Traugott Reference Hopper and Closs Traugott2003, Roberts & Roussou Reference Roberts and Roussou2003, Van Gelderen Reference Gelderen2004). Grammaticalization does not just occur abruptly; there is usually a gradual and partially overlapping shift from one form to another, with variation as to whether the source element remains available alongside the newly created functional item or not (see again the above-cited sources).
In their investigation into the grammaticalization of perception and causative verbs in Dutch, VCVK (2015, 2017) identify three imperative(-like) contexts where these verbs occur and they propose a three-step grammaticalization process or path. The three imperative(-like) contexts that they investigate are (i) regular imperatives (see example (5)), (ii) inflected imperatives, where the imperative verb displays number agreement with the direct object or with the subject of the embedded infinitival (see example (6)), and (iii) imperatives as discourse markers, where the imperative verb does not have a directive interpretation anymore (see example (7)).
VCVK (2015, 2017) recognize five criteria in the existing literature that can be used to indicate how the verbs have grammaticalized across these three contexts (Hopper & Traugott Reference Hopper and Closs Traugott2003). These criteria highlight the characteristics of functional elements as compared to lexical ones. Since grammaticalization entails the change of a lexical item into a functional one, these can therefore be seen as features probing the extent of grammaticalization. The five characteristics are (VCVK 2015:2):
According to these criteria, VCVK note that causative and perception verbs in regular imperatives have not undergone any grammaticalization: None of the above-mentioned features characterize these elements, which behave like regular lexical verbs. When used as discourse markers, these causative and perception verbs are the most grammaticalized: All of the above-mentioned features are characteristic of these elements. Finally, in inflected imperatives, they appear to be somewhere in-between on the grammaticalization path: Only some of the above-mentioned features apply to these elements.
Similarly, the PLI-constructions that are considered in this article are imperative-like structures with the causative verb laat/laten ‘let’ that no longer has an imperative meaning. According to Aikhenvald (Reference Aikhenvald2010:346) and Kuteva et al. (Reference Kuteva, Bernd Heine, Long, Narrog and Rhee2019:251), let is a common grammaticalization target. By drawing on the criteria used by VCVK and highlighting some other characteristics of PLI-constructions, I will show that PLI-laat/laten ‘let’ in Afrikaans and Dutch has undergone grammaticalization in a similar way to the perception and causative verbs in VCVK’s study.
2.1 Lack of argument structure
The first of VCVK’s criteria is that functional elements generally lack argument structure. VCVK (2015, 2017) assess this criterion by examining whether the relevant structures include an imperative pro-subject (external agent) or not. They use reflexive binding as a test, and note that in regular imperatives, there is a syntactically present second-person pro-subject which can bind a reflexive; agreeing imperatives as in (6), however, cannot contain a reflexive and therefore a pro-subject seems to be lacking in this case. In contrast to the transitive perception verb horen ‘hear’ that VCVK used when testing this criterion, let-imperatives do not have the kind of internal argument structure that can be reflexivized. Instead, let-imperatives feature first- and third-person causees, that is, indirect objects, which fit the general profile of let-type hortative and jussive constructions (Collins Reference Collins2004, Aikhenvald Reference Aikhenvald2010). Given these differences, it is necessary to first discuss the regular argument structure of let-imperatives before exploring the argument structure of PLI-laat/laten ‘let’.
The usual argument structure of let-imperatives is generally assumed to involve three components: a causer, which is the agent of laat ‘let’ and the addressee of the imperative; a caused event (e.g. going to the party, as in (1)); and a causee, which is the agent of the caused event and is represented by an oblique-marked first- or third-person embedded subject. This is illustrated abstractly in simplified form in (9).
In these constructions, the causer is typically either a null or nominative-marked second-person subject, aligning with the second-person subject found in imperatives more generally. While the subject is often null, as in (1) and (2), it can also be made overt by using the nominative form jy ‘you’, which is coreferential with the addressee of the imperative. This is shown in (10).
Assuming that the schematic representation in (9) illustrates the argument structure of ungrammaticalized laat ‘let’, we can now explore the extent to which this argument structure is maintained (or not) in the Afrikaans and Dutch PLI-constructions.
Let us first consider Afrikaans PLI-constructions. The hypothetical PLI allows for either an oblique-marked embedded subject or a nominative-marked one, as illustrated in (11).
Given this case alternation, I propose that Afrikaans in fact has two types of hypothetical PLI-constructions: a Type 1 hypothetical, as in (11a), and a Type 2 hypothetical, as in (11b).
In contrast to hypothetical PLIs, Afrikaans mirative PLI-constructions only allow nominative-marked subjects following laat ‘let’, as illustrated in (12).
Afrikaans Type 2 hypothetical and mirative PLIs therefore show a notable departure from the argument structure of regular let-imperatives. As illustrated above, these PLIs feature a nominative pronoun following laat ‘let’, which indicates that the pronoun no longer functions as the causee: Causees in let-imperatives are oblique-marked forms. This already suggests that such PLI-constructions do not have the regular argument structure of let-imperatives, as set out above (see (9)). Moreover, the identity of the causer – which pinpoints a specific entity for regular causatives – has become vague and unspecifiable in the case of Type 2 hypothetical and mirative PLIs: The interpreted causer corresponds to the kind of impersonal or generic subject found in impersonal constructions (Cinque Reference Cinque1988, Cabredo Hofherr Reference Hofherr, Ackema, Brandt, Schoorlemmer and Weerman2006). This shift is particularly evident in the fact that, unlike in regular let-imperatives, a nominative second person subject (jy ‘you’) cannot be added to these constructions.
Furthermore, again in contrast to let-imperatives, the nominative pronoun following laat ‘let’ in Type 2 hypothetical and mirative PLI-constructions is not limited to the second person; it can appear in various person and number combinations. This is illustrated with a first-person plural subject in (14a) and a third-person singular pronoun in (14b).
In contrast, the Type 1 hypothetical PLI still has an oblique-marked causee, and a second-person nominative pronoun (jy ‘you’) can be added between laat ‘let’ and the embedded subject, mirroring the structure of regular let-imperatives. This is shown in (15).
Only second-person nominative subjects are permitted in this position; first- and third-person nominative subjects are not allowed. This all suggests that, while Type 2 hypothetical and mirative PLIs have lost the argument structure of let-imperatives, Type 1 hypothetical PLIs have retained this grammatical structure.
Turning to Dutch, the hypothetical PLI permits only an oblique causee like mij ‘me’ (see again (4a)); a nominative pronoun ik ‘I’ cannot be used in this case, as shown in (16). This suggests that Dutch only has the hypothetical PLI form that has retained the usual argument structure of let-imperatives.
On the other hand, Dutch mirative PLI-constructions behave like Afrikaans hypothetical PLIs in that they can select either a nominative subject or an oblique causee, with no clear interpretational difference between the two forms. An example with the nominative pronoun (which I refer to as Type 2) is shown in (4b) above, while one with an oblique causee (Type 1) is given in (17).
The empirical facts outlined in this section reveal which PLI-constructions have retained the usual argument structure of let-imperatives and which have not. The constructions in which laat/laten ‘let’ takes a causee argument (e.g. my/mij ‘me’) do not lack argument structure; these mirror the regular imperative let-constructions in the types of arguments that they can take. This is illustrated in simplified form in (18), where laat ‘let’ is assumed to be a light verb (v) selecting a causer-subject (PRO), an embedded subject DP-causee (my), and a caused event-VP of which the causee is the agentive subject.
Conversely, the PLI-constructions in which laat/laten ‘let’ is followed by a nominative subject (ek/ik ‘I’) lack this regular causer–causee argument structure of let-imperatives. This deviation suggests that laat/laten in these cases are no longer functioning as regular light verbs (v).
In sum, both the Afrikaans Type 2 hypothetical and mirative PLI-constructions, as well as the Dutch Type 2 mirative PLI, cannot be represented as in (18), as they lack the argument structure of regular let-imperatives. In contrast, the Afrikaans Type 1 hypothetical, along with the Dutch hypothetical and Type 1 mirative PLI, can be represented as such, as they have retained the regular let argument structure and involve all three components assumed for let-imperatives: a causer, a causee, and a caused event.
2.2 Semantic bleaching
VCVK also investigate whether the imperative verb has undergone semantic bleaching. During the process of grammaticalization, it is common for the lexical meaning of an item to be weakened or even lost. At the same time, other meanings may also become associated with the item (Hopper & Traugott Reference Hopper and Closs Traugott2003:3). Von Fintel (Reference Fintel1995:185) suggests that, as part of this shift from lexical to more functional, a lexical item may become permutation-invariant, meaning that it loses specific reference to individual entities, attributes, or real-world situations.
In Afrikaans, both the Type 2 hypothetical and mirative PLIs have undergone semantic bleaching in this sense, as the permissive or causative meaning of laat ‘let’ has been lost or at least weakened. This semantic shift is especially clear in cases where the PLI-laat co-occurs with the regular causative laat, as in (19). In these examples, the boldfaced instance of laat ‘let’ contributes a causative interpretation, while the clause-initial PLI-laat instead serves a function parallelling that of the complementizers respectively indicated as alternatives to laat in the examples in (19) (see section 2.6 for further discussion of this pattern).
The Type 1 hypothetical PLI can also co-occur with another instance of laat ‘let’, as illustrated in (20).
In this case, however, both occurrences of laat convey a causative meaning, indicating that Afrikaans permits the stacking of causative elements within a single clause (see e.g. Nie Reference Nie2020).
In the analysis presented in de Villiers (Reference de Villiers2025: chapter 4), I interpret these examples as representing two distinct cases. In the Type 1 hypothetical PLI, the clause-initial laat functions as a regular causative v (as evidenced by its interpretation and argument structure), which can optionally co-occur with a further (lower) causative verb, and which moves to the clause-second (V2) position in C during the derivation. In contrast, in Type 2 hypothetical and mirative PLIs, the initial laat-forms lack both the permissive or causative interpretation and the regular argument structure of the light verb laat ‘let’; they have become permutation-invariant in the sense of von Fintel (Reference Fintel1995). As such, they are analyzed as grammaticalized C-elements, which can appear alongside a regular causative verb lower in the clause.
The same general observations apply to Dutch: PLI-constructions with an oblique-marked embedded subject (i.e. those that retain the let-imperative argument structure) still feature a permissive laten ‘let’. This is made clear by the translation of the hypothetical PLI in (4a), repeated here as (21a). In contrast, laten ‘let’ in Type 2 mirative PLIs with a nominative subject has been semantically bleached and can co-occur with a regular causative v, as illustrated in (21b).
2.3 Phonological reduction
The third of VCVK’s criteria deals with phonological reduction. As noted above, the possibility for an element to be phonologically reduced can serve as an indicator of grammaticalization. In Afrikaans, all three PLI-forms allow for laat ‘let’ (pronounced as /lɑːt/) to be phonologically reduced to lat (pronounced as /lʌt/), which entails the shortening of the vowel to /ʌ/. This is illustrated in (22).
This, however, is not a distinguishing feature of Afrikaans PLI-constructions as many forms of laat (/lɑːt/) can be reduced to lat (/lʌt/) in colloquial Afrikaans. Consider the exhortative let-imperative in (23).
That said, the patterns of phonological reduction between the PLI-forms do indicate a subtle distinction. Although a short vowel is possible, the laat-form in Type 1 hypothetical PLIs (i.e. the regular permissive v) is most typically pronounced with a long vowel (/lɑːt/). On the other hand, the Type 2 hypothetical PLI-laat can be pronounced with either a long or short vowel (/lɑːt/ or /lʌt/), while the mirative PLI-laat is most commonly pronounced with a short vowel. Thus, the reduced form /lʌt/ appears to be the default in mirative PLIs.
In standard Dutch, neither of the PLI-constructions allow for phonological reduction of laten ‘let’, as is illustrated in (24).Footnote 5
2.4 Lexical specialization
The fourth characteristic of functional items used by VCVK is that they are closed class items. VCVK use this criterion by testing whether all, some, or no other perception or causative verbs can be used in the imperative-like context that they are testing. What they are testing for is so-called lexical specialization: to the extent that the Dutch perception and causative verbs have become lexically specialized for use in a given context, they cannot be replaced by another perception or causative verb. They note that all perception and causative verbs can be used in regular imperatives, while only a select few of these verbs can be used as discourse markers, making them specialized for use in this latter context.
In Afrikaans, laat ‘let’ is lexically specialized for use in both the Type 2 hypothetical and mirative PLIs, as it cannot be replaced by the other Afrikaans causative verb maak ‘make’ (see (25a, b)). In contrast, although not completely natural, it is possible for the Type 1 hypothetical PLI-laat to be replaced by maak ‘make’, suggesting that laat is not lexically specialized for use in this case. This is illustrated in (25c).
Likewise, the Dutch hypothetical and Type 1 mirative PLI-laten ‘let’ can also be replaced with the causative verb doen ‘do’, as illustrated in (26a, b). While this substitution is also not entirely natural, it is not ungrammatical, indicating that laten ‘let’ is not fully specialized as the only causative verb that can be used in this case. On the other hand, replacing laten ‘let’ with doen ‘do’ in the Type 2 mirative PLI is completely out (see example (26c)), which means that laten ‘let’ is specialized for use in this context.
2.5 Morphological defectiveness
The last characteristic of functional items that VCVK consider is their morphological defectiveness: Functional items may lose some of their morphological properties during the process of grammaticalization. VCVK consider this criterion by testing whether the horen-imperatives they are probing have corresponding indicative forms. They observe that while regular imperatives have an indicative form, inflected imperatives and imperatives as discourse markers cannot be used indicatively. Here, I interpret this criterion as one that probes whether PLI-structures allow V2-forms, as we would expect for verbs that can be used indicatively, or whether they are restricted to V1-structures, which would not be expected for regular verbs.
A V2-structure is permitted in Afrikaans Type 1 hypothetical PLIs, as long as the second-person subject jy ‘you’ is overtly present. In such cases, the nominative second-person subject can occupy the initial position, as illustrated in (27a); (27b) shows that non-subject-initial V2-structures are also possible.
This pattern aligns with what we expect from a V1-conditional with an overt subject. In Type 1 hypothetical PLIs, laat functions as a regular permissive or causative verb that undergoes V2-movement to the C-position, thereby allowing an initial XP.
On the other hand, Afrikaans Type 2 hypothetical and mirative PLIs consistently exhibit V1-word order and cannot be V2. This is illustrated in (28).
Similar observations hold in the case of Dutch: The hypothetical and Type 1 mirative PLI (although the latter is more marked) do allow a V2-structure, while the Type 2 mirative PLI does not. This is illustrated on the basis of the hypothetical PLI in (29a) and the Type 2 mirative in (29b).
The lack of indicative counterparts (or V2-structures) in the Type 2 hypothetical and mirative PLIs in Afrikaans, as well as the Type 2 mirative in Dutch follows if the initial laat in these structures is no longer verbal, as was suggested in the preceding discussion. One consequence of this fact is that grammaticalized laat cannot participate in the movement to C that produces the standard V2 word order. Further, an indicative construction in non-null-subject languages like Afrikaans and Dutch differs from V1-imperatives in that it requires an overtly realized subject. The argument-structure considerations discussed in section 2.1 therefore become significant: in the absence of a syntactically represented agent-causer, it is not possible to create a well-formed indicative from Type 2 hypothetical and mirative PLIs in Afrikaans or from the Type 2 mirative in Dutch. By contrast, the laat/laten ‘let’ forms used in Afrikaans Type 1 hypothetical PLIs, and in Dutch hypothetical and Type 1 mirative PLIs still show remnants of the regular let-imperative argument structure, which means that they do have a syntactically present agent-causer. It is therefore possible for these structures to be realized in the indicative, with an overt subject.
2.6 Complementizer alternation
In addition to VCVK’s criteria, there are two other additional properties of these constructions that could add to the understanding of how they have grammaticalized. The first of these concerns complementizer alternation. In the introduction to this article, I highlighted the similarity between hypothetical PLIs and conditional structures. Given the interpretive parallel, it is reasonable to expect that the meaning conveyed by hypothetical PLIs could also be expressed using the conditional complementizer as ‘if’. This expectation is borne out: Both Afrikaans Type 1 and Type 2 hypothetical PLIs can occur with the complementizer as, as shown in (30).
However, the nature of the alternation between laat ‘let’ and as ‘if’ differs in the two constructions. In Type 1 hypothetical PLIs, as co-occurs with laat, resulting in a structure that retains both elements, as in (30a). In Type 2 hypothetical PLIs, on the other hand, as replaces laat entirely, which means that as and laat do not appear together in these cases; they are in complementary distribution; see (30b). Dutch hypothetical PLIs behave like the Type 1 hypothetical PLIs in Afrikaans, where laat and als ‘if’ can co-occur.
Afrikaans mirative laat, on the other hand, can alternate with insubordination structures introduced by the complementizer dat ‘that’, without a change in meaning (see example (31)). Insubordination is the main clause use of a structure that looks like a subordinate clause (Evans Reference Evans and Nicolaeva2007:367).
This insubordination option is not available in the Afrikaans hypothetical PLIs, nor is it possible in Dutch hypothetical or mirative PLIs.
These complementizer alternation facts once again support the proposal that Type 2 hypothetical and mirative PLI-laat are grammaticalized complementizer elements rather than light verbs. They appear in complementary distribution with the complementizers as ‘if’ and dat ‘that’.
2.7 Obligatory use of SAP-elements
The final property concerns the (non-)obligatory presence of speaker/hearer-oriented elements; henceforth sap-elements. When modal particles become obligatory in specific contexts, they shift from being purely optional expressive elements to obligatory functional elements which fulfil a crucial function in the syntax.
As illustrated by most of the examples discussed so far, modal particles such as nou, net, and wraggies (in Afrikaans) are typically present in PLI-constructions and contribute the speaker-oriented interpretation associated with these structures. In Afrikaans hypothetical PLIs, these SAP-elements are essential for conveying the speaker’s perspective – a hallmark of PLI-constructions. Omitting them still gives a grammatical conditional imperative, but while these forms remain structurally well-formed, as (32) shows, they no longer qualify as speaker-oriented PLI-constructions (see de Villiers Reference de Villiers2025 for a detailed discussion of the interpretation of these structures).
In contrast, Afrikaans mirative PLIs are not well-formed without SAP-elements and thus, require these elements to be present in the structure:
Similarly to Afrikaans hypothetical PLIs, Dutch Type 1 and Type 2 mirative PLI-constructions also require SAP-elements (like modal particles) to bring about the speaker-oriented interpretation. However, these elements can be omitted without rendering the sentence ungrammatical. The particle-less examples in (34) are, nevertheless, slightly marked.
In contrast to all other PLI-forms, the Dutch PLI-construction with the hypothetical interpretation seems to be preferred without the modal particles,Footnote 6 as in (35).
In sum, the above-mentioned facts suggest that, while most of the PLI-constructions considered here require SAP-elements to encode the structure’s speaker-oriented meaning, omitting them does not result in ungrammaticality. The only structure that is an exception here is the Afrikaans mirative PLI, which obligatorily requires an SAP-element. This indicates that, at least in this case, the PLI-structure is dependent on both the grammaticalized form of the verb laat ‘let’ and the activation of the grammatical structure associated with SAP-elements.
2.8 Data summary and analytical conclusions
Table 1 summarizes the facts considered above.
Table 1. Grammaticalization properties of Afrikaans and Dutch PLI-constructions
As illustrated in the table, PLI-constructions that select an oblique-marked embedded causee – such as Dutch hypothetical and Type 1 mirative PLIs, as well as Afrikaans Type 1 hypothetical PLIs – retain the regular argument structure of the permissive or causative verb laat/laten ‘let’. In contrast, PLI-constructions with a nominative embedded subject do not retain this causer–causee argument structure, indicating that these laat/laten-forms are no longer regular permissive or causative light verbs.
This structural divergence is mirrored in the semantic contributions of the laat/laten-forms. PLI-constructions that preserve the original argument structure (i.e. those that pattern with let-imperatives) still have the permissive/causative interpretation to some extent. Those that have lost this structure have also undergone semantic bleaching: laat/laten ‘let’ in these contexts no longer conveys the lexical meaning of the verb.
Furthermore, laat/laten ‘let’ in the PLI-constructions that mirror let-imperatives are not specialized for use in these cases, and can be replaced with other causative verbs, such as Afrikaans maak ‘make’ or Dutch doen ‘do’. This substitution, however, is not possible in PLIs that have lost the permissive/causative meaning and argument structure of laat/laten.
A further distinction emerges in terms of morphological defectiveness: the PLI-constructions that have retained the regular let argument structure can be used indicatively (i.e. with a V2-structure), due to the presence of a frontable verb and a causer-agent. In contrast, the more grammaticalized PLIs, which lack both the verbal element and the causer-agent argument, are restricted to V1-structures.
The above-mentioned properties of the different PLIs can be accounted for under the analysis outlined in de Villiers Reference de Villiers2025 (which can be consulted for a more detailed account). Since laat/laten in the first group of PLI-constructions (i.e. Dutch hypothetical and Type 1 mirative PLIs and Afrikaans Type 1 hypothetical PLIs) has retained the regular argument structure and the permissive/causative meaning associated with laat, I propose that these laat-forms are light verbs v non-distinct from the laat/laten in regular permissive structures. Conversely, since laat/laten in the second group of PLIs (i.e. Afrikaans Type 2 hypothetical and mirative PLIs and Dutch Type 2 mirative PLIs) no longer exhibits verbal properties – in interpretive, argument-structure and movement terms, I propose that these uses of laat/laten are not verbs but C-elements first merged in that position. This is illustrated on the basis of the Afrikaans Type 2 hypothetical in simplified form below.
All hypothetical PLI-laat/laten can alternate with the conditional complementizer as/als ‘if’, but the nature of this alternation varies. Only in Afrikaans Type 2 hypothetical PLIs can the complementizer as directly replace laat. In Dutch hypothetical and Afrikaans Type 1 hypothetical PLIs, laat/laten co-occurs with as/als, appearing in its first-merge position in vP. In contrast, only the Afrikaans mirative PLI allows alternation with the complementizer dat ‘that’, which is used in insubordination contexts. As with as in Afrikaans Type 2 hypothetical PLIs, dat and mirative PLI-laat occur in complementary distribution. These patterns further support the analysis: in cases where laat can be replaced by a complementizer, it functions as a C-element; in cases where it co-occurs with a complementizer, it functions as a light verb.
The obligatory use of SAP-elements further distinguishes the constructions under investigation. Except for Dutch hypothetical PLIs, all PLI-types require SAP-elements (e.g. modal particles) to bring about their characteristic speaker-oriented interpretations. However, only the Afrikaans mirative PLI requires these elements for grammaticality; their absence results in a sentence that is ill-formed. This suggests that the Afrikaans mirative PLI laat has grammaticalized beyond the C-domain to become an illocutionary complementizer (see Corr Reference Corr, Berns, Jacobs and Nouveau2018 for a detailed discussion of complementizers of this type). As a head within the SAP-domain, mirative laat plausibly requires the presence of another SAP-element (see de Villiers Reference de Villiers2025: chapter 4 for detailed discussion of the formal consequences of laat’s integration with the SAP-domain).
With the above in place, the observed phonological reduction patterns can also be adduced as additional evidence of the varying degrees of grammaticalization exhibited by the Afrikaans PLI-constructions. Recall that the permissive/causative laat, as used in Type 1 hypothetical PLIs, is generally pronounced with a long vowel (/lɑːt/); the laat-form used in Type 2 hypothetical PLIs, which is more grammaticalized than Type 1 hypothetical laat, may feature either a long or short vowel (/lɑːt/ or /lʌt/); and in mirative PLIs, the laat-form is typically pronounced with a short vowel (/lʌt/), aligning phonologically with dat ‘that’, and reflecting its advanced grammaticalized status.
In sum, the characteristics of the PLI-constructions reflected in Table 1 above indicate that both Dutch PLI-types are not as grammaticalized as their Afrikaans counterparts, and thus, that Afrikaans PLI-laat ‘let’ is more grammaticalized than Dutch PLI-laten ‘let’.
3. Internal factors, contact considerations, or both?
In the previous section, I built on VCVK’s (2015, 2017) study of the grammaticalization of Dutch perception and causative verbs to show that PLI-laat/laten ‘let’ has undergone a similar grammaticalization process. Going back to the main question of this special issue, it therefore seems like the development of the PLI-constructions in Afrikaans and Dutch was heavily influenced by a process that is often discussed as a grammar-internal process of change, namely, grammaticalization. Importantly, however, the existence of contact-induced grammaticalization (see Heine & Kuteva Reference Heine2003) is also well-established, with more recent work (e.g. Walkden & Breitbarth Reference Walkden and Breitbarth2019) arguing that the nature of a contact situation will determine the speed at which grammaticalization processes progress. In the following, I will briefly consider how these ideas may be relevant to the development of the Afrikaans and Dutch PLI-constructions.
Afrikaans has been in contact with numerous languages from different language families throughout its history (see among others Ponelis Reference Ponelis1993, Roberge Reference Roberge1994, Deumert Reference Deumert2004, Van der Wouden Reference van der Wouden2012). It was deeply influenced by Dutch, but other European settlers (like the French, German, English, and Portuguese), the indigenous Khoe-people, and slaves from other African and Asian countries also played their part in the development of Afrikaans. Furthermore, Afrikaans developed primarily as a spoken language which needed to be used, initially frequently by L1-speakers of languages it was in contact with, within a multilingual society (Roberge Reference Roberge, Deumert and Vandernbussche2003). Importantly, its subsequent development also took place in a sociolinguistically complex context.
Biberauer (Reference Biberauer2018) highlights the significance of speaker–/hearer-perspectival elements and structures in contact languages. These typically develop as spoken languages, and, accordingly, feature a range of in part highly productive devices (designated lexical items and syntactic structures) to convey speaker-hearer-related (i.e. (inter)subjective) meanings. Afrikaans is no different in this regard: in fact, given the fact that it has always been shaped and used in sociolinguistically complex situations, throughout its history, it is unsurprising that both standard Afrikaans and particularly the modern spoken varieties feature numerous quite clearly intersubjectively oriented elements. Being primarily used to convey the speaker’s perspective – the speaker either sketches their perspective on some or other hypothetical event, or they express their surprise/shock about a certain fact in the here-and-now – the PLI-constructions discussed in this paper instantiate intersubjectively oriented elements par excellence.
Language contact can lead to language change on all linguistic levels, depending on the extent to which contact has occurred and how similar the relevant languages are, among other things (Gooskens et al. Reference Gooskens, van Bezooijen, Kürschner, Norde, de Jonge and Hasselblatt2010, Trudgill Reference Trudgill2011). According to Trudgill (Reference Trudgill2011), the set-up of different contact situations can produce quite different results, with both simplification and complexification of the contact language being possible outcomes: In short-term contact situations with many second language learners, simplification is expected; in long-term contact situations with many bilinguals, by contrast, complexification is expected. Building on the view that there is a link between language structure and the sociolinguistic environment it developed in, Walkden & Breitbarth (Reference Walkden and Breitbarth2019) claim that the speed at which language changes like grammaticalization can occur is influenced by contact. A well-studied instance of language change is the Jespersen Cycle. It has been argued that language contact can affect progression within this cycle (Beyer 2009, Lucas & Lash Reference Lucas and Lash2010) and also the speed at which a language moves from one stage to the next (Rutten et al. Reference Rutten, van der Wal, Nobels and Simons2012, Breitbarth Reference Breitbarth2014, Walkden & Breitbarth Reference Walkden and Breitbarth2019). More specifically, in cases of intense, short-term language contact, it appears that languages move to the next stage faster.
Returning to Afrikaans, we might then expect faster progression within a given grammaticalization process (here: the upwards reanalysis of laat from light verb v to complementizer C) on account of the intense contact that has taken place throughout its development. On the other hand, it is important to emphasise the often overlooked fact that Dutch-derived varieties were spoken as mother-tongue languages at the Cape and over a far longer period than is often appreciated (see, among others, the work of Hans den Besten, Christo van Rensburg, and Frank Hendricks); that is, the short-term language-contact component of Trudgill’s proposal does not apply in the case of Afrikaans as bi- and multilingualism involving Afrikaans has been in place for a long time. This contact factor might therefore be expected to interact with the intensity one just highlighted.
As the discussion in section 2 has shown, the Afrikaans PLI-structures do indeed appear to have developed at a faster rate than their Dutch counterparts. It is therefore tempting to appeal to the proposals in Trudgill (Reference Trudgill2011) and Walkden & Breitbarth (Reference Walkden and Breitbarth2019). The intensity-of-contact component of these proposals predicts the faster development of Afrikaans PLIs. Crucially, though, the short-contact/L2-learner component of these models does not apply as obviously, leaving open the question of the precise role of contact in the observed developments. Taking the PLI-structure’s fundamentally interactionally-oriented nature into account, we conclude that the currently unclear/potentially spurious dividing line between grammaticalization (understood as the creation of further functional material) and pragmaticalization (understood as the creation of specifically discourse- and interaction-oriented functional material; see Müller & Axel-Tober Reference Müller and Axel-Tober2025) may be key in probing the question more fully. At this point, however, it seems reasonable to ascribe a role to both internal and external factors in accounting for the respective rise and development of PLI-constructions in Afrikaans and Dutch.
4. Conclusion
In this article, I have shown that the PLI-laat/laten ‘let’ has undergone grammaticalization in Afrikaans and Dutch. I illustrated this by drawing on the grammaticalization criteria used by Van Craenenbroeck & Van Koppen (2015, 2017) in their work on perception and causative verbs in Dutch. I have shown that the development of the PLI constructions is consistent with a process often viewed as ‘internal’. Furthermore, it is a well-established fact that imperative-like let often undergoes grammaticalization (see Aikhenvald Reference Aikhenvald2010 and Kuteva et al. Reference Kuteva, Bernd Heine, Long, Narrog and Rhee2019). In addition to the grammaticalization facts, I have also shown that Afrikaans PLIs appear to have grammaticalized further than their Dutch counterparts, and I have suggested that language contact may have played a role in the speed at which these constructions developed in the two languages. Therefore, both internal changes and language contact may have played a role in the development of PLI constructions in Afrikaans and Dutch.
Open access
