The uses of clitic si in Child Italian

Received: 19-06-2025 Accepted: 15-09-2025 Published: 29-11-2025 Universitat Autònoma de Barcelona The uses of clitic si in Child Italian Chiara Dal Farra University of Milano-Bicocca 0000-0001-9226-5644 [email protected] Silvia Silleresi University of Milano-Bicocca 0000-0001-6733-8733 [email protected] Artemis Alexiadou ZAS Berlin & Humboldt-Universität zu Berlin 0000-0002-6790-232X [email protected] Uli Sauerland ZAS Berlin 0000-0003-2175-535X [email protected] Maria Teresa Guasti University of Milano-Bicocca 0000-0002-4470-4187 mariateresa.gua[email protected]t Abstract This paper investigates the different uses of the clitic si in Child Italian. Through a corpus study on spontaneous productions of children aged 1;4-3;4, we check whether How to cite: Dal Farra, Chiara, Silleresi, Silvia, Alexiadou, Artemis, Sauerland, Uli & Maria Teresa Guasti. 2025. The uses of clitic si in Child. Isogloss. Open Journal of Romance Linguistics 11(1)/12, 1-28. DOI: https://doi.org/10.5565/rev/isogloss.597 2 Isogloss 2025, 11(1)/12 Dal Farra et al. all functions of si are realized by children and we give a new perspective from language acquisition to a long-standing debate, namely whether there exists only one type of si or whether there are many different ones. Our results show that Italian children use si productively and adult-like early on, and they produce all its different functions: impersonal, anticausative, true reflexive, and inherent reflexive – although at different rates. We claim that our results support the one si approaches, and the difference in the frequencies of si functions reflects the structural complexity of some structures. Keywords: clitic si, Italian, first language acquisition, impersonal, syntax. 1. Introduction The clitic si in Italian appears in a wide variety of contexts, and despite extensive theoretical research its role and interpretation are still heavily debated. In this paper, we focus on some of its uses, observing them through the lenses of child language acquisition. We discuss the result of a corpus study, determining the types and order in which the different functions of si emerge in children’s speech. This could help us contributing from a new perspective to the long-standing debate between competing sets of theories that have been proposed in the literature (for an overview see D’Alessandro 2007, Pescarini 2015): one stating that all types of si can be reduced to the same element which covers all the different functions, and the other claiming that there exist (at least) two types of sis which are (syntactically and/or semantically) unrelated to each other. The paper is structured as follows. Section 2 provides a brief overview of the Italian uses of si in adult language, illustrates current theoretical accounts for an analysis of the clitic si, and finally summarizes the state of the art on the acquisition of this clitic in Romance languages. Section 3 presents the research questions and hypothesis, and Section 4 illustrates our study. In Section 5 the results of the corpus study are presented. Section 6 discusses the main findings, while Section 7 concludes the paper. 2. The different uses of si in Italian In Italian, as in all Romance languages, the clitic si (or se in Catalan, French, Portuguese, Romanian, and Spanish) can appear in several contexts assuming different functions. The main uses of this clitic in Italian can be listed as follows: true reflexive (1), reciprocal (2), inherent reflexive (3), anticausative (4), impersonal (5 - 6), passive (7), and middle (8). 1 1 The list is not exhaustive, as other uses have been individuated and discussed in the literature. See the case of the so-called (low) applicative si, as in (i), mentioned in D’Alessandro (2007) and analyzed in Campanini and Schäfer (2011) and in Martin and Arunachalam (2022). (i) Livia si è bevuta un frullato. Livia si be.3SG drunk a smoothie ‘Livia drank a smoothie.’ Clitic si in Child Italian Isogloss 2025, 11(1)/12 3 (1) Maria si pettina. Maria si comb.3SG ‘Maria combs herself.’ (2) Le maestre si abbracciano. the teachers si hug.3PL ‘The teachers hug each other.’ (3) Sebastiano si addormenta. Sebastiano si fall.asleep.3SG ‘Sebastiano falls asleep.’ (4) Il vaso si è rotto. the vase si be.3SG broken ‘The vase broke.’ (5) In quel ristorante si spende molto. in that restaurant si spend.3SG much ‘In that restaurant one/people/they spend(s) a lot.’ (6) Domani si va al mare. tomorrow si go.3SG to.the sea ‘Tomorrow we are going to the sea.’ (7) In Italia si mangiano gli spaghetti. in Italy si eat.3PL the spaghetti ‘In Italy people eat spaghetti.’ (8) Questo libro si legge facilmente. this book si read.3SG easily ‘This book reads easily.’ In true reflexive si, as in (1), the two theta-roles are assigned to the same entity. In Cinque (1988)’s terms, reflexive si absorbs the external theta-role and accusative case, since it is merged as the external argument of the verb and undergoes head movement towards I. For this reason, the internal argument moves to the subject position and receives nominative case (see also Burzio 1986, D’Alessandro 2007). What yields the reflexive interpretation is either the co-indexation between the moved DP (internal argument) and si (see Belletti 2020 for a detailed explanation), or a bundling operation (in terms of Reinhart and Siloni 2005). Reflexive si can take on another value as in (2), namely the reciprocal one. Reflexives and reciprocals have the same distribution: they must refer to, and be bound by, a subject antecedent. These are traditionally considered bound anaphors, hence subject to principle A of the Binding Theory (Chomsky 1981, Lebeaux 1983). However, they differ in that reciprocals must have a plural interpretation (see Manzini 1986). Given that two arguments are needed for these constructions, the reflexive and reciprocal uses are allowed only with transitive verbs. 4 Isogloss 2025, 11(1)/12 Dal Farra et al. Inherent reflexives, as in (3), are found with a restricted set of verbs, i.e. unaccusatives. That is to say, verbs without an external argument – and consequently which cannot assign accusative case. 2 For such a reason, in some analyses, si operates only in the lexicon, remaining syntactically inactive (Reinhart and Siloni 2005): it does not have a role associated (Burzio 1986, Cinque 1988, Rizzi 1986) and, specifically, it cannot bear an agentive construction, since the external argument is absent. Anticausative si – also called unaccusative or ergative reflexive – is found with causative verbs which undergo the causative alternation between transitive and intransitive use. With some verbs, the intransitive use is obligatorily marked with the clitic, as in the case of rompere ‘break’ in (4). 3 In many analyses, anticausative si marks suppression or lack of the external argument and of accusative case, but it does not have an implicit role associated, i.e. it is syntactically inactive and is attached to the verb in the lexicon rather than in the syntax (Cinque 1988, Manzini 1986, Reinhart and Siloni 2005). In some other cases it is analyzed as a semantic expletive: given the lack of an external theta-role, si is not a thematic argument of these verbs, but is still considered syntactically active, in that the clitic is first merged in an A-position (Schäfer 2017). One function of si that has received much attention in Italian is the impersonal one, where even though the subject is not specified, it can nonetheless be understood and identified – depending on the construction – as ‘people/they/one/we’. Its interpretation is arbitrary, being associated not only with a third-person feature, but also with unspecified number and gender features (Cinque 1988, Manzini 1986). The reference set selected by si can be either purely generic, as in (5), or there may be a group of people satisfying the property expressed by the predicate, the so-called existential reading. The generic reading often (but not exclusively) gets an arbitrary reading roughly corresponding to ‘people’ or ‘they/one’. The existential reading may be further specified for inclusiveness (speaker-inclusive reading), or it may not. 4 The speaker-inclusive reading, as in (6), is considered productive mainly in central varieties of Italian - Tuscan dialects in particular (D’Alessandro 2007). It basically corresponds to a first person plural ‘we’ and can be considered as the most unrestricted reading, since it contains all the participants in a context (on this point see Cinque 1988, Chierchia 1995, Holmberg and Roberts 2013). Impersonal si can be used with all verb types: it can be either the agent of transitive and unergative verbs, or the theme/patient of unaccusative verbs (Cinque 1988). In previous approaches, it has been analyzed in different ways: (i) as a clitic in the I node, which governs an empty NP (pro) in subject position and hence receives the theta-role associated with the subject, as well as nominative case (Belletti 1982, Cinque 1988, Kayne 1986); (ii) as a clitic subject that arises via movement creating a chain with an empty category (Burzio 1986); (iii) as a free variable generated in the syntax, which must be bound to an empty category (expletive) in the subject position and form a chain with it, and is 2 Notice that in some approaches, it is even considered parallel to an anticausative structure (Cennamo 2014, Folli 2002). 3 Although notice that there are different classes of verbs in this respect: those which are obligatory marked, those that are not (affondare ‘sink’), and those that are optionally found with the clitic (bruciare ‘burn’). 4 See D’Alessandro (2007: 137) for cases in which the speaker is excluded from the reading of si: Mi si è raccontato che Maria ha riso molto ieri ‘I have been told Maria laughed a lot yesterday’. Clitic si in Child Italian Isogloss 2025, 11(1)/12 5 hence associated with nominative case (Manzini 1986). Whatever derivation we align with, impersonal si seems to differ from all the other si functions in that it must be bound to the subject and is associated with nominative case. Another widely discussed function is the passive (passivizing or passivizer) si. In such uses, si – which can only occur with transitive verbs – absorbs the external theta-role. The verb agrees with the internal argument, which can stay in the base position or can be moved to the pre-verbal one, in a derivation that is similar to what we find in traditional passives (Belletti 1982, Burzio 1986, Hyams 1983). However, unlike what is found in passives, note that (9) and (10) do not license any by-phrase (see Pescarini 2015 for a detailed description of other differences between passive si and traditional passive). A fairly common view in the literature is that passive si occurs in sentences where the subject and verb agree with each other (Belletti 1982, Cennamo 2014, Lepschy 1986), as exemplified in (9) and (10) 5 . (9) In Italia si mangiano gli spaghetti (*da tutti). in Italy si eat.3PL the spaghetti (by everyone) ‘In Italy people eat spaghetti.’ (10) In Italia si mangia la pizza (*da tutti). in Italy si eat.3SG the pizza (by everyone) ‘In Italy people eat pizza.’ However, in cases of singular agreement it is not possible to distinguish between an impersonal and a passive use – whereas it is in case the agreement is plural, as in (9). According to Lepschy (1986), in fact, a sentence like the one in (10) can be paraphrased as ‘the pizza is eaten’ or ‘one eats pizza’, while the sentence in (9) can only receive a passive interpretation, e.g. ‘the spaghetti are eaten’. Other approaches embrace a different analysis, where both structures (9) and (10) with plural and singular agreement are considered instances of impersonal generic constructions (Cinque 1988, D’Alessandro 2007, Pescarini 2015). Recent proposals on analogous Spanish constructions suggest that, despite what agreement facts might suggest, in these constructions the internal argument of the verb is not the subject but the direct object throughout the derivation, while si is the active nominative pronominal subject (see Ormazabal and Romero 2019, 2024 for a detailed analysis). Finally, in some of the formal literature (Cinque 1988, Pescarini 2015), the term ‘middle si’ is used to identify a particular construction apparently identical to the impersonal/passive one in which, however, any specific temporal reference is absent. This is exemplified by the case in (8). Typically, middle si occurs with modal adverbs and with transitive verbs. It has been proposed that si blocks, or suspends, the external argument and accusative case. Consequently, the internal argument moves to the preverbal position (Cinque 1988, Manzini 1986). In other words, middle si can be 5 A variant for (9) is the sentence in (i) where the agreement on the verb is singular. Although we acknowledge the existence of such cases, they will not be discussed in the present work, since they are not acceptable to all speakers and seem to undergo idiolectal variation (D’Alessandro 2007). (i) In Italia si mangia (gli) spaghetti. in Italy si eat.3SG (the) spaghetti ‘In Italy people eat spaghetti.’ 6 Isogloss 2025, 11(1)/12 Dal Farra et al. considered as an instance of the impersonal one (Manzini 1986, Cinque 1988 – although see Pescarini 2015 for a different conclusion). To further reinforce this idea, notice that the agreement patterns of middle si are not different from those of impersonal/passive si with the possibility of having plural verb-internal argument agreement (D’Alessandro 2007). In sum, the list of si functions can be grouped as follows: (i) true reflexive (1) and reciprocal (2); (ii) inherent reflexive (3); (iii) anticausative (4); and (iv) impersonal. Impersonal si includes the generic (5) and speaker-inclusive (6) readings, but it also covers other functions that could be seen as instances of impersonal constructions – in particular of impersonal generic constructions: passive (7), and middle (8). In conflating these functions, we follow some of the accounts presented so far (Cinque 1988, D’Alessandro 2007, Manzini 1986). In all the uses listed above, the presence of si is mandatory: we can say that the clitic si serves as a marker to get to those meanings and, apparently, has a general function of absorbing or blocking part of the argument structure (generally the external argument) of the verb. We will return to this point in Section 6. However, a question remains: is it merely coincidental that such seemingly different constructions rely on the same clitic form? This question has sparked significant debate within generative grammar from the 1980s to the present day: the ‘one si vs. many sis’ debate (see Pescarini 2015 for an overview). The following paragraph briefly outlines the theories supporting these two approaches. 2.1. The one si versus many sis debate Before moving on to our study, it is worth mentioning the main approaches to si that have been proposed in the literature. As anticipated, previous accounts fall into two main set of theories: (i) ‘many sis’ theories, claiming the existence of (at least) two different sis (Belletti 1982, Burzio 1986, D’Alessandro 2007, Napoli 1973, Pescarini 2015, a.o.) and (ii) ‘one si’ theories (Belletti 2020, Cinque 1988, Kayne 1986, Manzini 1986, Manzini et al. 2016) which claim (with different degrees of abstraction and generalization) for the existence of only one si common to all – or to the majority of – functions. Within the first set of theories, different divisions among si-types have been proposed. The first attempt dates back to Napoli (1973), who proposes that si has two functions: a reflexive one and an insertion one. The former governs reflexives, reciprocals, anticausatives, and middles, while the latter applies to impersonal and medio-passive structures. These two functions are considered semantically and syntactically distinct. Burzio (1986) distinguishes several types of sis. First, a reflexive one, which is base-generated in clitic position and forms a chain with an empty category in object position that bares both case (accusative) and theta-role (object). Another type marks unaccusativity, signaling the lack of assignment of a thematic-role to the subject, but without a syntactic role: it is simply an affix. In his analysis, this is the case for inherent reflexives as well as for anticausatives. And finally, a si type dedicated to impersonals and passives, classified as a subject clitic associated with nominative case, which must be cliticized by movement. Along the lines of Burzio’s argument, D’Alessandro (2007) and Pescarini (2015) isolate the impersonal si as distinct from the other sis, being characterized by Clitic si in Child Italian Isogloss 2025, 11(1)/12 7 the fact that it introduces an unspecified subject in the clause and can co-occur with other si functions (notably the reflexive one) occupying a lower position in the structure. Depending on the analysis, impersonal si also covers the function of middle and passive si (D’Alessandro 2007) or only of passive si, being a different variant – with plural verb agreement – of the si impersonal construct (Pescarini 2015). Belletti (1982) proposes an even more subtle distinction between impersonalactive and impersonal-passive si, which are claimed to undergo different syntactic processes: while the former receives nominative case and external theta-role, the latter is assigned accusative case and external theta-role. Functioning as a passive morpheme, si can absorb the accusative case the verb would typically assign to its direct object. Additionally, in this role, si also absorbs the external theta-role. This allows for a passive configuration whereby the object moves to the subject position and is assigned nominative case. This idea has been further developed in Belletti (2020) via the means of acquisition data: she theorizes the existence of a reflexivepassive si that may represent a possible intermediate step to get to traditional passive constructions in Italian child-language. The derivation proposed for this reflexivepassive function is the same as in Belletti (1982) for passive si, with the exception that si is co-indexed with the DP subject, yielding to a reflexive interpretation in the adult grammar, but not necessarily triggered in child production. Belletti suggests that this analysis aligns with the unified approach of the various types of si in Italian. Given the fact that she reduces the impersonal construction in Italian to an instance of passive construction (Belletti 1982), and given that in tensed clauses middle si can always be interpreted as an instance of impersonal-passive si (Cinque 1988), Belletti suggests that these structures could be considered as an extension of what she calls reflexivepassive si. The first hypothesis proposing a unified nature of si dates back to Kayne (1986), who suggested that the underlying mechanism shared by all functions consists of the lack of assignment (or withholding) of the thematic role to the external argument. This generates an unaccusative structure in which the internal argument, if present, moves to the subject position and receives case. This proposal was further developed a few years later by Cinque (1988), who speculated that all uses of si might follow from the same basis, with different levels of specifications attached to it. He distinguished between argumental [+ arg] and non-argumental [- arg] si, where only the former can absorb external theta-role. True reflexive and reciprocal si are argumental, whereas both inherent reflexive and anticausative sis are non-argumental: they are syntactically inactive (and, in fact, are found only with verbs which do not assign external thematic role) and are basically the result of lexical processes. Impersonal si can either be argumental or non-argumental, depending on the verbs it occurs with: if the verb assigns an external theta-role it will be argumental, whereas non-argumental si is found with all verb classes. Finally, passive si is non-argumental. The unification can be obtained at a certain level of abstraction: (i) impersonal [± arg], (ii) passive [-arg] and (iii) reflexive (true, reciprocal, inherent and anticausative). [± Arg] sis respectively (i) absorb [+arg]/identify [-arg], (ii) suspend [-arg] and (iii) absorb [+arg]/ suspend [-arg] the external theta-role. In other words, all functions of si somehow block part of the argument structure of the verb. Manzini (1986) offers one of the first unified accounts for si that takes into account both syntactic and semantic properties of each function. She manages to unify impersonal and reflexive si on syntactic bases: both sis are bound by the subject (chain 8 Isogloss 2025, 11(1)/12 Dal Farra et al. formation relation or referential dependence relation), the only difference being that impersonal si forms a chain with the subject, while reflexive si is in a referential dependence relation with it. On the semantic side, they both have a (free or dependent) variable-like character: all uses develop from the same lexical item from which different functions can arise depending on (i) its association with a freeor a dependent-variable, and (ii) the presence or absence of a passivizer property, which blocks the assignment of the external theta-role in the subject position. These are summarized in Table 1. Table 1. Manzini’s derivation of impersonal, reflexive and middle sis. variable passivizer impersonal free - reflexive dependent - middle free + Source: Adapted from Manzini (1986: 259). Impersonal si is a free variable, where the passivizer property is not realized. On the other hand, when the passivizer property is realized with a free-variable, we are dealing with a middle si. Reflexive si is a dependent variable, since it must be bound by the subject, and no passivizer property is realized. 6 Notice, however, that in this account one major category is left out: anticausative si is considered to operate in the lexicon, where it attaches to the verb. Hence, it works differently from the sis described here, which instead play a role in the syntax. Manzini (1986) assumes that in such a case its semantic properties are meaningless (see also the proposal in Schäfer, 2017). Eventually, Manzini et al. (2016) unifies anticausative constructions with the other functions, suggesting that in this case si is a free variable that cannot interpret the external argument altogether (which, in this approach, is projected). Despite extensive theoretical debate and several theories that have attempted to define unified or multiple derivations of the clitic si, no studies exist in the acquisition domain that investigate its various uses in a way that could inform one theory or another. We think that acquisition data can contribute to the theoretical debate on the existence of one or more types of si. For this reason, in the next section we focus on the few studies that investigated the acquisition of the clitic si/se in Romance languages. Furthermore, we briefly present the few studies on Italian child language that investigated the acquisition of some uses of the clitic si. 2.2. Si/Se in Child language To our knowledge, to date there are no studies considering the acquisition of all the different uses of si in Italian. Hence, studies on other Romance languages regarding the acquisition of this clitic will be briefly discussed in this section. We will report the results of some studies conducted in Peninsular Spanish and French. The uses of se (Italian si) in these languages reflect only partially those found in Italian, given that impersonal – either generic or speaker-inclusive – is expressed with a different clitic 6 In Manzini’s proposal another function is taken into account, namely a middle-reflexive use, which is a dependent variable with passivizer property, as exemplified here: gli unici bambini lavatisi ‘the only children (who) washed themselves’. Since such cases were not produced by children in our study, we excluded them from Table 1. Clitic si in Child Italian Isogloss 2025, 11(1)/12 9 in French (on) and is not reported in acquisition studies in Spanish. We will conclude this section reporting on the very few studies that looked at the acquisition of some functions of si in Italian. As for Peninsular Spanish, in a corpus study based on CHILDES Teomiro Garcia and Escobar Alvarez (2013) looked for productions and omissions of se in one child (age range 1;9-4;7). They showed that anticausative and inherent reflexive ses were produced adult-like since the age of 2 years. Although impersonal se is an option in Spanish, it was not mentioned in their study. Pierce (1992) tested the acquisition of Spanish middle-passive se constructions in 18 children aged 3;7-5;9 via a semiimitation elicited production task. The results showed that children were more likely to produce sentences with the object in post-verbal position (V-S passive) than in preverbal one (S-V passive). Performance on the se-passives with pre-verbal object improved with age and became fully productive only at 5-6. According to the author, the outcome of the facilitated production of V-S passives when compared to S-V passives was due to a lack of A-chains, as nominative case may be assigned under government to the subject in post-verbal position, but not in pre-verbal one (similar to Ormazabal and Romero 2024’s analysis). Regarding the acquisition of clitic se in French, different studies have looked at its approximate age of acquisition, yielding contradictory results. Heinen and Kadow (2010) tried to determine the phase of acquisition of different French morphosyntactic constructions on the basis of an analysis of diary studies of 18 children and indicated the acquisition of reflexive se between 2;1-3;8. Similarly, Hamann et al. (1996) analyzed a longitudinal corpus collected on one subject and found that reflexive se was produced by the child only around 2;6. The largest study on this topic is the one by Barrière et al. (2000) and the following work Barrière and Lorch (2006), who searched and identified all se-constructions in large corpora of speech production (3 diary studies, 2 CHILDES corpora, 2 cross sectional corpora of children aged between 2-4 and 6-7) collected on 200 children. Data analysis revealed that reflexive and reciprocal ses were the first to appear (already at 2;1) and that anticausative/middle-passive se constructions appeared very shortly after (from 2;5). To sum up, previous studies suggest that (i) anticausative and inherent reflexive se are acquired first in Spanish, (ii) middle-passive se in Spanish is fully acquired around the age of 5-6, and (iii) reflexive se in French is acquired slightly before anticausative and middle-passive uses. However, these studies do not take into account the use of impersonals, which would be possible only in Spanish, since French relies on a different clitic (on) to realize this function. The only acquisition data we have on the impersonal function is presented in Hyams (1983), where she reported the emergence of impersonal si constructions with singular agreement in Italian children aged 2;5-2;9, followed shortly thereafter by those with plural agreement at 2;10. From these data, it seems that impersonal si in Italian appears at the same age as the other functions in Spanish and French. However, Hyams did not detail the frequency with which impersonals were produced, nor did she include a comparison with other si constructions. Our study aims to fill this gap. Although the findings presented here suggest an early acquisition of reflexive se in Spanish and French, there are conflicting results in the literature regarding Italian children’s language development. It has been widely assumed that by 3-4 years of age children have command of Principle A, know what a reflexive is, and know that it must be locally bound in a variety of languages, including Italian (for an overview see 16 Isogloss 2025, 11(1)/12 Dal Farra et al. Table 4. Verb types per si functions in Child Italian. Raw numbers (percentages). Impersonal Anticausative Inherent reflexive True reflexive Transitive 166 (90.71%) - - 24 (100%) Unaccusative 11 (6.01%) - 16 (100%) - Unergative 3 (1.64%) - - - Alternating - 82 (100%) - - Modal 3 (1.64%) - - - 5.2. Interim discussion When examining our results, we observe that all si functions were produced and appeared around the same time. However, we need to account for the overwhelming use of impersonal si with respect to the other si functions. Before moving on to a detailed analysis of this group in the next section, we wanted to make sure that the lower production of the other three functions – true reflexives, inherent reflexives and anticausatives – with respect to impersonal si was an accurate reflection of the corpus and not due to (i) cases of omission of the clitic – which were not considered in our first search in the corpora – and (ii) a bias of the conversational context. Concerning (i), we looked for omissions of clitic si in true reflexive, inherent reflexive, and anticausative uses. 13 To do this, we conducted a corpus search with the four verbs most frequently used with these functions, namely mettere ‘put’, fare male ‘hurt’, nascondere ‘hide’, vestire ‘dress’ – true reflexives; sedersi ‘sit down’, scottarsi ‘burn’, alzarsi ‘get up’, arrabbiarsi ‘get angry’ – inherent reflexives; and chiamare ‘call’, rompere ‘break’, sporcare ‘dirt’, bagnare ‘wet’ – anticausatives. Children rarely omitted the clitic si: only 11/98 times. These were distributed as follows: 3 cases of omissions in true reflexives, 4 in inherent reflexives, and 4 in anticausatives. The fact that Italian children do not omit si in anticausative constructions has also been independently shown in Silleresi et al. (2024). to answer this question. However, we still consider the use of si as being productive, and we think that the uses of ‘si + verb’ are not formulaic given the following points: (i) there is a great variety of verbs used, (ii) children use si with lexical verbs and with modal verbs, and (iii) they also use the same verb with and without the clitic. Moreover, as for the latter case, we performed another corpus search in every file of 3 children (Camilla (Antelmi), Diana, and Guglielmo (Calambrone)) and we looked for the 8 verbs most frequently attested with si. We noticed that not only were these children using both the singular and the plural form, but also that the same child in the same file uttered the verb using different inflections, showing that they have performed a morphological analysis of the verbs as claimed in Guasti (1993/1994). In any case, the crucial point for us is that the utterances of ‘si + verb’ involve a total 73 different verbs. So, even if some of these verbs combined with si are indeed more frequent in adult language, we still find a large variety of verbs in the children’s speech. Moreover, it should also be stated that some of these verbs (that is, bare verbs, not their occurrence with the clitic si) are simply frequent in the vocabulary of young Italian children, as in the case of mettere (put) or fare (do). For these reasons, we would still like to claim that the use of si indeed is productive. 13 We did not take into account possible omissions of si in impersonal cases, as these are impossible to be spotted: in fact, if si is omitted in this function, in Italian it basically corresponds to a pro-drop sentence. Clitic si in Child Italian Isogloss 2025, 11(1)/12 17 Regarding (ii), since the recorded sessions reported in the CHILDES corpora consisted of one-to-one interactions between the child and an adult (parent or experimenter), we wanted to make sure whether these contexts were more prone to elicit 1st (mi) and 2nd (ti) person reflexive clitics, rather than 3rd person ones. In other words, we checked whether the paucity of reflexives in our results was due to a bias of the conversational context. For this reason, we searched for the number of mi and ti reflexive (both true and inherent) occurrences. As a result, we found that 1st and 2nd person reflexive clitics were mi = 16 occurrences; ti = 4 occurrences for true reflexives; mi = 13 occurrences; ti = 5 occurrences for inherent reflexives. Putting them together with si, we obtain a total of n = 44 occurrences of true reflexives and n = 34 occurrences of inherent reflexives. This seems to indicate that the lower frequency of reflexive si was not entirely affected by a conversational bias. We can then conclude that reflexive si (both true and inherent) was less produced with respect to other functions and that anticausatives were less produced than impersonal. We will speculate on why this was the case in the discussion. The fact that impersonal was the most frequently and first produced function constitutes a novelty in the acquisition literature. Let us now focus on describing how this function was realized by Italian children. 5.3. A focus on impersonal (middle-passive) si First, we separated the occurrences of impersonals between the two subtypes evoked in Section 2: generic and speaker-inclusive readings. Among impersonal constructions, 98 had a generic reading (53.55%) and 33 a speaker-inclusive one (18.04%). The remaining 52 occurrences were ambiguous (28.41%) between the two readings. 14 Concerning the type of verb used with impersonal constructions, they were distributed as follows: generic si = 92 transitives (93.87%), 3 modals (3.06%), 2 unaccusatives (2.05%) and 1 unergative (1.02%); speaker-inclusive si = 23 transitives (69.69%), 9 unaccusatives (27.27%) and 1 unergative (3.04%). In sum, the majority of impersonal utterances produced by children were constructions with transitive verbs. As explained in Section 2 and 4, we grouped passive and middle si with instances of impersonal generic si. Furthermore, as we know, passives and middles are restricted to instances of transitive verbs. Hence, in order to individuate possible cases of passive and middle si, we searched for these among all cases of generic impersonals with transitive verbs, examining three properties: (i) verb agreement (singular vs. plural) between the verb and the DP, as in (16), in relation to (ii) the presence and position of the internal argument (pre-verbal DP vs. post-verbal DP vs. pro-drop), as in (17), and (iii) presence of an adverb. These properties would be an indicator of whether children already have some of the machinery that would lead to the production of instances of passive (properties (i) and (ii)) and middle (properties (ii) and (iii), and possibly (i)) si. In particular, for passive si we could expect the possibility to move the 14 These cases could not be disambiguated via the context. Thus, the interpretation remained ambiguous between a generic (people/one) and a speaker inclusive (we) reading (e.g. e si porta via ‘and si takes away’, Calambrone: Diana 2;6). 18 Isogloss 2025, 11(1)/12 Dal Farra et al. internal argument to the pre-verbal position, 15 as well as plural agreement on the verb. In the case of middle si we should find the internal argument in the pre-verbal position, as well as the use of modal adverbs. Table 5 summarizes these results. (16) a. (Tonelli: Marco 2;5) Si batte qua. si beat.3SG here ‘You have to beat here.’ b. (Calambrone: Guglielmo 2;9) ma quelli lì si mettono così. but those there si put.3PLP like.that ‘But those over there should be placed like this.’ (17) a. (Antelmi; Camilla 3;4) Quello non si mangiava […] that not si eat.3SG ‘That one could not be eaten…’ b. (Calambrone: Diana 2;6) e poi si mette qui la tazza. and then si put.3SG here the cup ‘And then you put the cup here/the cup should be put here.’ c. (Tonelli: Marco 2;5) così (pro) si tira so (pro) si pull.3SG ‘You pull it like this.’ Table 5. Properties of DPs in impersonal/middle-passive with transitive verbs (raw numbers). We did not find any occurrence of modal adverbs, suggesting that (unambiguous) middle si was not yet produced by children. Concerning agreement on the verb, almost the totality of the occurrences displayed singular agreement, whereas only 5/92 occurrences had plural agreement. 16 Among the former, DPs were mostly 15 Although notice that, as pointed out by an anonymous reviewer and as we stated in Section 2, the pre-verbal or post-verbal position of the internal argument is not the best diagnostic, given that for both periphrastic passives and si passive either options are allowed. Nonetheless, it has been shown that there is a difference with respect to the preand postverbal position in children’s uses, as well as in how soon they master them. Specifically, Pierce (1992) showed that the post-verbal position is acquired earlier and used more often than the pre-verbal one in Spanish middle-passive se constructions. Hence, we can speculate that we could also have seen a difference in our data, and in particular that the use of the pre-verbal position would more likely be linked to a passive interpretation than the post-verbal one. However, given the fact that numbers were low in general, and that the rate of DPs in both positions was similar, we could not draw any conclusion. 16 The few instances of plural agreement were spread in almost the whole dataset starting from 2;1. Agreement verb and DP: Singular Agreement Plural Agreement DP: PreV PostV Drop PreV PostV Drop 15 16 56 2 1 2 Clitic si in Child Italian Isogloss 2025, 11(1)/12 19 omitted, while when realized they were equally distributed in preand post-verbal position. This does not allow us to draw any conclusions with respect to property (ii). Concerning the latter, not much can be said except that the shortage of plural agreement suggests that impersonal with plural agreement (unambiguous passive) si was almost never produced by children. Taken together, these results seem to indicate that it is not entirely possible to distinguish between the functions of impersonal, passive, and/or middle si in most of our data coded as generic impersonal constructions. In fact, as detailed in Section 2 and 4, instances of si with singular agreement (and most with the drop of DPs) fall into a grey area in which it is not entirely possible to assign a function over the other. This group hence might just be better referred to as “impersonal/middle-passive”. 6. Discussion The research presented here contributes to our understanding of how the clitic si is acquired in Italian. The data we provide have implications not only for the study of the relative order in which different si functions are acquired, but also for the broader theoretical framework concerning the nature of si – whether it is a single element that covers all the available functions, or whether there are instead multiple distinct sis which differ substantially, with little or nothing to do with each other. Our results indicate that Italian children used si constructions in a productive and adult-like way from a very early age (1;8-3;4). Furthermore, they were able to correctly associate verb types and si functions. Impersonal (middle-passive) constructions were the most frequently and the earliest produced, followed by anticausative constructions and true and inherent reflexives. Anticausatives emerged around the same time as impersonals (1;8), while reflexives appeared slightly later (1;11-2;1). Among the functions of impersonal si, both speaker-inclusive and generic si were produced. In the latter, we notice that it is hard to assign an unambiguous function between impersonal and middle-passive, since si with a modal adverb (unambiguous middle) was not yet realized by Italian children at the age of 3;4, whereas impersonals with plural agreement (unambiguous passives) were rarely produced (see also Hyams 1983 for similar results). The results of our study hence seem to align with the one si hypothesis, as all functions of si – impersonal (middle-passive), anticausative, true reflexive and inherent reflexive – appear within a few months in children’s speech. In fact, if there were issues with the acquisition of this polyfunctional clitic, we would expect children to struggle for an extended period of time in the production of multiple si functions, as suggested in the literature on the acquisition of homophones (Anderson 1998, Clark 1991, Doherty 2004, Gavarró 2020, Mazzocco 1997, Prévost 2009, Storkel and Maekawa 2005). However, this was not the case. These results were further confirmed by the fact that children rarely omitted the clitic si and that they produced it adult-like in almost the whole dataset. That said, some aspects of the data still require further explanations, as certain functions were more produced than others. Notably, there was a higher frequency of impersonal (middle-passive) constructions compared to other functions of si. A first possible explanation is that impersonal si can be used with a wider range of verb types (transitive, unergative, unaccusative, modal) which is not the case for all 20 Isogloss 2025, 11(1)/12 Dal Farra et al. other si functions (Cinque 1988). This explanation works well for the cases of inherent and anticausative si: as they occur with a limited set of verbs (unaccusatives in the case of inherent reflexives, and verbs that undergo the causative alternation in anticausatives), we can expect these functions to be quantitatively less produced than impersonal si, as indeed we found in our data – especially in the case of the inherent reflexive. Anticausative si, despite being less produced than impersonal si, was highly attested in the dataset. We speculate that the higher presence of anticausative si with respect to true and inherent reflexive si may depend on how frequent these functions are in adult Italian. To understand whether this is a possible explanation, we would need to check the distribution of the clitic si in the adult language. Unfortunately, the only data existing in this respect rely on a very different classification of si functions (Topciu and Chesi 2019). 17 Since this paper focused on the acquisition of the clitic and the timing of production of its functions in child Italian, rather than their distribution in the language, we leave this aspect open for future research. However, the fact that impersonal si is used with all verb types does not explain why, even when we limit the comparison to transitive verbs, these constructions were still produced at a much higher rate than other constructions that require the same verb type, notably true reflexive. The same can be said for those functions of impersonal si restricted to transitive verbs, namely impersonal with plural agreement (unambiguous passive) and instances with an adverb (unambiguous middle). This discrepancy suggests that factors beyond verb type may contribute to the higher frequency of impersonal si. We argue that the derivation of impersonal si is syntactically less complex than those of other si functions. This may explain why children produced it more frequently. Let us explore why this might be the case, starting from the comparison between impersonal and reflexive. We posit – as previously detailed in Section 2 – that impersonal si is a clitic that must be bound to the subject. In this scenario, si licenses the presence of a null pro subject (following Rizzi 1986), or an expletive pro (following Cinque 1988). It then forms a chain with pro, through which it receives nominative case. On the other hand, in true reflexives and in reciprocals, si must absorb the external argument and take accusative case, then the internal argument must move to the subject position, and si must be anaphorically bound with it (Belletti 2020). When comparing the two derivations, it becomes clear that reflexives require some additional machinery beyond what is needed for impersonal si. It is not then surprising that children find it easier to produce impersonals compared to reflexives, and that the errors that have been found in previous literature in Italian are linked to the lack of binding of the moved internal argument with the clitic si (Belletti and Manetti 2019, Belletti 2020, Raminelli and Belletti 2021). 17 In fact, Topciu & Chesi (2019) check whether a purely distributional approach used in an automatic classification is able to correctly identify the type of si. Their classification is based on the position of the argument to which si is co-referent to, and it includes the following categories: (i) impersonal, (ii) local with the co-referent DP immediately preceding, (iii) local with the co-referent DP immediately preceding and modified by a PP or a relative clause, (iv) co-referent DP is post-verbal, and (v) the referent is absent and not retrievable from the context (i.e., mainly pro-drop cases). Such classification was not telling to us, given that the position of the co-referent DP is actually overlapping in different si functions, as in the case of reflexives (both true and inherent) and anticausatives. Clitic si in Child Italian Isogloss 2025, 11(1)/12 21 Now, let us move to the comparison between impersonal si and the functions of impersonal si restricted to transitive verbs. As we saw from our results, instances of unambiguous passives were rarely produced by Italian children, and middles were even absent form children’s productions. When we compare the derivation of these functions with the impersonal one, it becomes clear that the former require some additional machinery beyond what is needed for impersonal si. In impersonal with plural agreement (unambiguous passives), the fact that the verb has to agree with the plural internal argument – which, depending on the analysis, can either become the subject in a derivation similar to the one of traditional passive (as in Belletti 1982), or it can remain the object of the sentence while si acts as the nominative pronominal subject (as in Ormazabal and Romero 2019, 2024) – in addition to the fact that si has to block the external argument, constitutes a more complex operation than the one we hypothesized for impersonal si. This idea is supported by our results, in that children prefer to produce the default singular agreement in almost the totality of the cases. Moreover, it has been independently shown that Italian-speaking children do not productively use third person plural agreement before 2;6 (Caprin and Guasti 2009). Similarly, in middle constructions, si blocks the assignment of the external theta-role and triggers movement of the internal argument to the subject position, much alike to what we find in the derivation of traditional passives (Manzini 1986). Given that the operation underlying traditional passive is acquired in Italian around 3;6-4;0 (Guasti 2017, Manetti 2013, Manetti and Belletti 2015, a.m.o.), it is not surprising that children do not produce unambiguous cases of passive and middle functions at 3;4. However, our results seem to suggest that, although the syntactic derivation of these functions is not fully acquired yet, children already have access to the basic operations that can lead to them. Notably, they can move the internal argument to the pre-verbal position and they can produce plural agreement on the verb. In addition, they know that these operations can only be performed with transitive verbs. In sum, we could argue that they seem to use impersonal as a function to lead the way for passive/middle constructions. Thus, these observations support the approaches for which the impersonal, passive and middle si are not really distinct from each other (Cinque 1988, D’Alessandro 2007, Manzini 1986). One independent cross-linguistic observation seems to further reinforce this idea: in some languages where the passive constructions are not available, e.g. Ewe (a Niger-Congo language of the Kwa group), the passive meaning is realized via the use of impersonal constructions, as shown in (18). (18) Ewe (Abigail Anne Bimpeh p.c.) wó tu aʃe aɖe-wó. 3PL build houses INDEF-3PL ‘They built some houses (lit.) = some houses were built’ In sum, we showed that impersonal si is syntactically the simplest derivation among all the functions of si. It is not surprising, then, that it is the most produced function by very young children in Italian. From an acquisition perspective, these findings seem to support a unified theory of si, in that multiple functions appear and are used at the same time. Furthermore, we suggest that the limited use or even absence of certain types of impersonal si is due to the need for children to acquire more complex derivations, which are only fully mastered as they grow older. 22 Isogloss 2025, 11(1)/12 Dal Farra et al. As a final remark, let us briefly focus on how unified si can be derived. To do so, we follow Manzini (1986)’s and Manzini et al. (2016)’s approaches. Hence, let us posit that si is an open (unsaturated) variable and is generated in the syntax as a clitic on a verb. As already discussed in Section 2.1, such a variable can be free or dependent and has a semantic role: that of saturating one argument. It can correspond to the interpretation of the external argument or the internal one of a transitive verb, to the external argument of an unergative verb, or to the internal argument of an unaccusative verb (Manzini et al. 2005). These interpretations are supplied at the LF interface, as shown in (19). (19) Adapted from Manzini et al. (2016: 141 (46)) [y [V x]] i existential/generic closure → (∃/Gen(eric)y, short passive/impersonal) ii agent = theme → (y = x, true reflexive, reciprocals) iii agent not interpreted → (anticausative) In the case of transitive verbs, if the external argument variable is quantificationally closed through existential or generic closure (19i), impersonal, passive, and middle sis are derived. Alternatively, the variable can be reflexivized by identifying the internal and external argument (19ii). In this case, true reflexive and reciprocals are obtained. Finally, if the variable is not interpreted altogether (19iii), we obtain the anticausative reading. One function is still not captured under this approach, namely the inherent reflexive. Two basic proposals can be made here: one way to obtain the correct result is not generating the external argument y at all. Following this reasoning, the clitic is hence not generated in the syntax but only in the lexicon (Burzio 1986, Cinque 1988), meaning that this function indeed behaves differently from the others. The second one is that the argument is somehow generated but is not interpreted. Given the intricacy of the matter, we leave this question to future research. 7. Conclusions In this paper we focused on the use of the clitic si in young Italian children through a corpus study. To our knowledge, this is the first study to investigate the acquisition and development of all functions of si in Italian. Our analysis suffers the limitation intrinsic to corpus searches: the number of children and the age range were limited, moreover contexts could not always help to disambiguate the function. Thus, our conclusions should be further explored via the use of experimental data, eliciting the production and comprehension of the different functions of si, even with older children. Nevertheless, our data show that children aged 1;8 to 3;4 use si productively and adult-like from a very early age and that they produce all its different functions (albeit at different rates): true and inherent reflexive, anticausative, and impersonal/middle-passive. Furthermore, the results suggest that some aspects of more complex structures – such as passive and middle si – have yet to be fully acquired. We claim that our findings support the view that si is the same element across all its different functions (in line with previous accounts by Manzini (1986) and Manzini et al. (2016)), and its interaction with principles of grammar produce its different uses or interpretations. Clitic si in Child Italian Isogloss 2025, 11(1)/12 23 Acknowledgments The authors wish to thank Adriana Belletti, Fabienne Martin, Maria Rita Manzini, Yining Nie, Diego Pescarini, and Florian Schäfer for valuable discussion and feedback in various stages of this work. We also wish to thank three anonymous reviewers for insightful comments and suggestions on the previous version of the manuscript. Finally, we thank all members of the LeibnizDream project, as well as the audiences of Going Romance 2021, IGG47, GALA 2022, and the workshop “Clitici da tutti (o quasi) i punti di vista”, held at the University of Milano-Bicocca, for fruitful comments. Chiara Dal Farra and Silvia Silleresi conceived the experimental questions, performed the corpus search, coded the data, and drafted the manuscript. Artemis Alexiadou, Maria Teresa Guasti and Uli Sauerland conceived experimental questions, participated in results discussions, gave feedback on the manuscript, and provided fundings. All authors approve the final version of the article. This project has received funding from the European Research Council (ERC) under the European Union’s Horizon 2020 Research and Innovation Programme (LeibnizDream, Grant Agreement No. 856421). Data availability All materials, data, and statistical analysis scripts are available at https://doi.org/10.5281/zenodo.17592802. References Anderson, Raquel T. 1998. The use of reflexive constructions by Spanish-speaking children: Differences across functions. Applied Psycholinguistics 19(3): 489-512. https://doi.org/10.1017/S0142716400010298 Barrière, Isabelle, Lorch, Marjorie, & Marie Thérèse Le Normand. 2000. Argument structure alternation in French children’s speech. In M. Perkins, & S. Howard (eds.), New Directions in Language Development and Disorders 139-147. Boston, MA: Springer. https://doi.org/10.1007/978-1-4615-4157-8_14 Barrière, Isabelle, & Marjorie Perlman Lorch. 2006. On the acquisition of ambiguous valency-marking morphemes”. In V. Torrens, & L. Escobar (eds.), The Acquisition of Syntax in Romance Languages 41: 23-49. Amsterdam: John Benjamins. https://doi.org/10.1075/lald.41.03bar Bates, Douglas, Mächler, Martin, Bolker, Ben, & Steve Walker. 2015. Fitting Linear Mixed-Effects Models Using lme4. Journal of Statistical Software 67(1): 1-48. doi:10.18637/jss.v067.i01. Belletti, Adriana. 1982. Morphological passive and pro-drop: The impersonal construction in Italian. Journal of Linguistic Research 2(4): 1-34. 24 Isogloss 2025, 11(1)/12 Dal Farra et al. Belletti, Adriana. 2020. (Reflexive) Si as a route to passive in Italian. In L. Franco, & P. Lorusso (eds.), Linguistic Variation: Structure and Interpretation, 73-86. Berlin, Boston: De Gruyter Mouton. https://doi.org/10.1515/9781501505201-006 Belletti, Adriana, & Claudia Manetti. 2019. Topics and passives in Italian-speaking children and adults. Language Acquisition 26(2): 153-182. https://doi.org/10.1080/10489223.2018.1508465 Brown, Roger. 1973. A first language: The early stages. Cambridge, MA and London, England: Harvard University Press. https://doi.org/10.4159/harvard.9780674732469 Burzio, Luigi. 1986. Italian syntax: A government-binding approach. Dordrecht: Springer. https://doi.org/10.1007/978-94-009-4522-7 Campanini, Cinzia, & Florian Schäfer. 2011. Optional se-constructions in Romance: Syntactic encoding of conceptual information. Hand-out for a talk delivered at GLOW 2011. Caprin, Claudia, & Maria Teresa Guasti. 2009. The acquisition of morphosyntax in Italian: A cross-sectional study. Applied Psycholinguistics 30(1): 23-52. https://doi.org/10.1017/S0142716408090024 Cennamo, Michela 2014. Passive and impersonal reflexives in the Italian dialects: Synchronic and diachronic aspects. In P. Benincà, A. Ledgeway, & N. Vincent (eds.), Diachrony and Dialects: Grammatical Change in the Dialects of Italy, 71-95. https://doi.org/10.1093/acprof:oso/9780198701781.003.0004 Chierchia, Gennaro. 1995. The variability of impersonal subjects. In E. Bach, E. Jelinek, A. Kratzer, & B. H. Partee (eds.), Quantification in Natural Languages, 107143. Dordrecht: Kluwer Academic Publishing. Chomsky, Noam. 1981. Lectures on government and binding. Berlin, Boston: De Gruyter Mouton. Chondrogianni, Vicky. 2007. Acquiring clitics and determiners in child L2 Modern Greek. Selected Papers on Theoretical and Applied Linguistics 17(2): 356-366. https://doi.org/10.26262/istal.v17i2.5569 Cinque, Guglielmo. 1988. On si constructions and the theory of arb. Linguistic Inquiry, 19(4): 521-581. Cipriani, Paola, Pfanner, Lucia, Chilosi, Anna Maria, Cittadoni, Lorena, Ciuti, Alessandro, Maccari, Anna, Pantano, Natalia, Pfanner, Lucia, Poli, Paola, Sarno, Stefania, Bottari, Piero, Cappelli, Giuseppe, Colombo, C., & Edy Veneziano. 1989. Protocolli diagnostici e terapeutici nello sviluppo e nella patologia del linguaggio. Protocollo 1/84. Ministero della Salute della Repubblica Italiana: Stella Maris Foundation. Clitic si in Child Italian Isogloss 2025, 11(1)/12 25 Clark, Eve V. 1991. Acquisitional principles in lexical development. In S.A. Gelman, & J.P. Byrnes (eds.), Perspectives on language and thought: Interrelations in development, 31-71. Cambridge: Cambridge University Press. https://doi.org/10.1017/CBO9780511983689.003 D’Alessandro, Roberta. 2007. Impersonal “si” constructions. Berlin, New York: De Gruyter Mouton. D’Odorico, Laura, & Stefania Carubbi. 2003. Prosodic characteristics of early multiword utterances in Italian children. First Language 23(1): 97-116. https://doi.org/10.1177/0142723703023001005 Doherty, Martin. J. 2004. Children’s difficulty in learning homonyms. Journal of Child Language, 31(1): 203-214. https://doi.org/10.1017/S030500090300583X Folli, Raffaella. 2002. Resultatives: Small clauses or complex VPs? In C. Beyssade, R. Bok-Bennema, F. Drijkoningen, & P. Monachesi (eds.), Romance Languages and Linguistic Theory 2000: Selected Papers from ‘Going Romance’ 2000, Utrecht 30 November-2 December, 153–170. Amsterdam: John Benjamins. https://doi.org/10.1075/cilt.232.10fol?locatt=mode:legacy Gavarró, Anna. 2020. The acquisition of clitics in the Romance languages. In Oxford Research Encyclopedia of Linguistics. https://doi.org/10.1093/acrefore/9780199384655.013.656 Guasti, Maria Teresa. 2017. Language Acquisition: The Growth of Grammar. Cambridge, MA: MIT press. Guasti, Maria Teresa. 1993/1994. Verb syntax in Italian Child Grammar: Finite and nonfinite verbs. Language Acquisition 3(1): 1-40. https://doi.org/10.1207/s15327817la0301_1 Hamann, Cornelia, Rizzi, Luigi, & Uli H. Frauenfelder. 1996. On the acquisition of subject and object clitics in French. In H. Clahsen (ed.), Language Acquisition and Language Disorders 14: 309-334. https://doi.org/10.1075/lald.14.13ham?locatt=mode:legacy Heinen, Sabine, & Helga Kadow. 2010. The acquisition of French by monolingual children. A review of the literature. In J. M. Meisel (ed.), Two First Languages, 4772. Berlin, New York: De Gruyter Mouton. https://doi.org/10.1515/9783110846065.47 Holmberg, Anders, & Ian Roberts. 2013. The syntax–morphology relation. Lingua, 130: 111-131. Hsieh, Li, Leonard, Laurence B., & Lori Swanson. 1999. Some differences between English plural noun inflections and third singular verb inflections in the input: The