On Partial Null Subject languages: Why pro-drop in Brazilian Portuguese and Russian became similar but not identical

Rev. Estud. Ling., Belo Horizonte, v. 30, n. 4, p. 1896-1935, 2022. eISSN: 2237-2083 DOI: 10.17851/2237-2083.30.4.1896-1935 On Partial Null Subject Languages: Why Pro-Drop in Brazilian Portuguese and Russian Became Similar But Not Identical Sobre as línguas de Sujeito Parcialmente Nulo: Por que o pro-drop em português brasileiro e em russo se tornaram semelhantes, mas não idênticos Nerea Madariaga University of the Basque Country (UPV/EHU), Vitoria-Gasteiz / Spain [email protected] https://orcid.org/0000-0003-4732-4144 Abstract: In this paper, I claim that a parametric view on change in pro-drop does not contradict the fact that not all the Partial Null Subject (PNS) languages display identical properties. I show that the contingent nature of diachronic change is the reason for the slight differences between PNS languages. Modern Russian (MR) and Brazilian Portuguese (BP) are two PNS languages that developed from Consistent Null Subject antecessors (Old Russian and European Portuguese) independently from each other. I account for the change in pro-drop experienced by these two languages, analyzing the properties usually related to the null subject parameter (verbal inflection, clitics, null objects, embedded and arbitrary null subjects), and show that the final parametric setting in both MR and BP was almost identical, with small differences that can be attributed to the different initial conditions for the change. Keywords: Pro-drop; null subjects; null objects; clitics; Consistent Null Subject languages; Partial Null Subject languages; Brazilian Portuguese; European Portuguese; Modern Russian; Old Russian. Resumo: Neste artigo, afirmo que uma visão paramétrica sobre a mudança em pro-drop não está em conflito com o fato de que nem todas as línguas de sujeito nulo parcial (PNS) apresentam propriedades idênticas. Mostro que a natureza contingente da mudança diacrônica é a razão para as pequenas diferenças entre as línguas PNS. O russo moderno Rev. Estud. Ling., Belo Horizonte, v. 30, n. 4, p. 1896-1935, 2022. 1897 (MR) e o português brasileiro (BP) são duas línguas PNS que se desenvolveram a partir dos antecessores de sujeito nulo consistente (antigo russo e português europeu) independentemente uma da outra. Relato a mudança de pro-drop experimentada por essas duas línguas, analisando as propriedades geralmente relacionadas ao parâmetro de sujeito nulo (inflexão verbal, clíticos, objetos nulos, sujeitos nulos subordinados e arbitrários), e mostro que a configuração paramétrica final em ambas MR e BP foi quase idêntica, com pequenas diferenças que podem ser atribuídas às diferentes condições iniciais para a mudança. Palavras-chave: Pro-drop; sujeitos nulos; objetos nulos; clíticos; línguas de sujeito nulo consistente; línguas de sujeito nulo parcial; português brasileiro; português europeu; russo moderno; russo antigo. Recebido em 16 de fevereiro de 2022 Aceito em 28 de março de 2022 1 Initial remarks A recurrent topic in the literature in recent years is the discussion about the nature of so-called Partial Null Subject (PNS) languages, since Holmberg (2005), Holmberg; Nayudu; Sheehan (2009), Biberauer; Roberts; Holmberg; Sheehan (2010), etc. On the one hand, there seem to be salient similarities between the languages making up this group, but at the same time, not all the features characterizing PNS languages are common to all of them. This paper aims to contribute to a more accurate characterization of PNS languages, through the analysis of the current status and historical development of Null Subjects (NSs) in two Indo-European languages, Brazilian Portuguese and Russian, which shifted from a Consistent Null Subject (CNS) pattern into a PNS pattern independently from each other. The comparison of the diachronic development of these two languages will shed light on the reasons why PNS languages are different and similar at the same time. The similarities, it will be argued, stem from commonalities in parameter settings, while the differences will be proven to be a consequence of the contingent circumstances that surround change each time in each language (LIGHTFOOT, 1999). 1898 Rev. Estud. Ling., Belo Horizonte, v. 30, n. 4, p. 1896-1935, 2022. In Section 2, I introduce PNS languages and some basic assumptions. In Section 3, I compare the pro-drop system in Brazilian Portuguese (BP) and Modern Russian (MR), pointing out their differences with respect to European Portuguese (EP) and Old Russian (OR), respectively.1 In Section 4, I show that the emergence of the PNS status in these two languages correlates with the emergence of restrictions in licensing NSs, and explain why that happened. In BP, the old CNS system decayed with the “weakening” of the person inflectional paradigm, triggered by the reorganization of the pronominal system (cf., i.a., DuARtE, 1993, 2000; MODESTO, 2000; NUNES, 2011). In Middle Russian, however, the breaking point was the loss of V-to-t movement, due to the whole rearrangement of the verbal system around new aspectual distinctions, which replaced the old tense distinctions (JUNG, 2018; MADARIAGA, 2022a, 2022b). 2 Consistent Null Subject (CNS) languages and Partial Null Subject (PNS) languages In recent years, the classical view on the so-called pro-drop parameter has been called into question after observing that its clustering effects, described by, i.a., Rizzi (1982), are less regular across languages than predicted by the standard parametric theory (cf. NEwMEyER 2005). To give an example, the correlation between the existence of NSs and the morphological richness of verbal agreement does not hold in a straightforward way. In addition, “morphological richness” (as opposed to “morphological poverty”) has been shown to be difficult to characterize in a precise way (cf. BIBERAuER, 2008 for an overview). Recently, Roberts (2010); Biberauer; Roberts; Holmberg; Sheehan (2010); Holmberg; Roberts (2014); and subsequent work, have revisited the properties of pro-drop and formulated new hypotheses about its parametric options. These authors emphasize the fact that PNS languages are especially heterogeneous, which in principle could be considered a problem if we want to propose a common parameter setting for all of them. 1 the change in pro-drop in Russian took place in the 16th century (between early and late Middle Russian), so I include early Middle Russian under the denomination OR, and late Middle Russian in MR. As for the synchronic description of BP, I follow the convention of Portuguese / Brazilian scholars (BARBOSA; DuARtE; KAtO, 2005; DuARtE; SILVA, 2016. i.a.) and consider EP a comparable antecessor of BP. Rev. Estud. Ling., Belo Horizonte, v. 30, n. 4, p. 1896-1935, 2022. 1899 However, in this paper, I will claim that the heterogeneous nature of PNS languages is not a problem for parametric setting. More specifically, I will show that, if two unrelated languages adopt the same parametric setting of pro-drop, the basic pro-drop properties will be essentially the same in both languages. However, some clustering properties, which are contingent on previous stages of the language, can differ between the two languages. In the following pages, I will prove this hypothesis by comparing the change in pro-drop experienced by two unrelated PNS languages, BP and MR. First, however, let us introduce some basic facts and assumptions about CNS and PNS languages, implied in the change analyzed in this paper. In CNS languages (e.g., Romance languages, such as Spanish, European Portuguese, and Italian), non-emphatic, non-stressed, noncontrastive pronoun subjects are typically dropped; cf. Peninsular Spanish (1a-b), while stressed, focused, or contrastive subjects must be overtly realized (1c): (Peninsular Spanish) (1) a. – ¿Cómo (*tú) quieres (*tú) la sopa? h ow (*you) want (*you) the soup – (*Yo) la quiero caliente. (*I) cl want hot ‘– How do you want the soup? – I want it hot.’ b. – ¿Ha pasado Juan por casa? – Sí, (*él) vino y se llevó el ordenador. yes (*he) came and SE took the computer ‘– Did Juan come home? – Yes, he came and took the computer away.’ c. – ¿Se ha llevado Juan el ordenador? – No, se lo ha llevado María, *(él) sólo lo ha mirado. not cl cl aux taken Maria *(he) just cl aux looked ‘– Did Juan take the computer? – No, Maria took it away; he just took a look at it.’ PNS languages, on the other hand, form a heterogeneous group and are not easy to define. In general, the baseline realization of pronominal subjects in PSN languages is overt, but they can be dropped under certain conditions. Depending on the person or tense/mood, PNS languages vary their pro-drop pattern. For example, in Hebrew, Standard Finnish, Jakaltek, and Kenga, 1st and 2nd person NSs are available, as opposed to 3rd person (cf. VAINIKKA; LEVy, 1999), while the reverse 1900 Rev. Estud. Ling., Belo Horizonte, v. 30, n. 4, p. 1896-1935, 2022. pattern, i.e., availability of 3rd person NSs, as opposed to 1st and 2nd person, is found in Shipibo, historical varieties of Germanic, Dinka, and Lamani (CAMACHO; ELíAS-ULLOA, 2010; wALKDEN, 2012). the first contrast is illustrated for Finnish in (2a-b): (2) a. (Sinä) puhut englantia. (you) speak.2sg English ‘You speak English’ b.*(Hän) puhuu englantia. ((s)he) speaks.3sg English Intended: ‘She/he speaks English.’(Finnish - Holmberg, 2005, p. 539) In this paper, I will follow Holmberg (2005, 2010), Holmberg; Nayudu; Sheehan (2009), Roberts (2010), and subsequent work, in that the relevant parametric option for setting a CNS versus a PNS language is the acquisition of a [+D] feature on T. I will also follow these authors by assuming that NSs are deficient minimal φPs with unvalued interpretable φ-features (φP[iφ:_]), which can in principle receive or not a referential value, and whose behavior varies according to the specific parametric setting on T in the language. The available structures for PNS versus CNS languages are described in detail in Madariaga (2022a, 2022b), and summarized here: (i) In CNS languages, T is endowed with a [+D] feature, and a NS (a deficient φP), located at [Spec,t], automatically receives a referential interpretation by D-matching with T. In these languages, any Topic at CP (null or overt) is able to allow identifying the specific reference of the NS; more specifically, the referential index of a null or overt topic at [Spec,CP] is copied by the unvalued D-feature of T, and, then, through Agree, by the NS, which at the same time matches its unvalued φ-features. The D-feature of T has no morphological expression of its own, and is “spelled out” by realizing person and number features, resulting in rich verbal morphology (HOLMBERG; NAyuDu; SHEEHAN, 2009). (3) [CP Topici C [tP φPi TD:_ [VP …]]] In the Spanish examples (1a-b) above, referential NSs are naturally licensed by virtue of the D-feature on T and their topical nature, while a [+focus] or [+contrastive] feature at CP forces the overt realization of the subject (1c). Rev. Estud. Ling., Belo Horizonte, v. 30, n. 4, p. 1896-1935, 2022. 1901 (ii) In PNS languages, t lacks a D-feature, so that a NS (a deficient φP) is able to match its φ-features with t, but does not necessarily raise to [Spec,TP], and does not automatically receive a referential index through T. In that case, the NS gets a generic or arbitrary interpretation. However, there can be some independent mechanism in the language, which is able to endow the deficient φP with a referential index (HOLMBERG; NAyuDu; SHEEHAN, 2009). In languages like Finnish, MR, and BP, there seem to be at least two alternative mechanisms to achieve this referentiality: (a) the first mechanism is related to pragmatics: NSs can receive a referential interpretation by copying the index of some null topic at C. This null topic is part of the discourse common ground, shared by the speaker and the hearer, i.e., it can convey either deictic (situational / contextual) features or logophoric features (for the 1st and 2nd person), which allow the reference of the NS to be recovered (SIGuRÐSSON, 2011; tSEDRyK, 2015). Depending on the particular conditions of the specific PNS language, it can be the case that referential NSs are licensed only in one or two grammatical person(s), or that all persons are licit in different pragmatic situations (tSEDRyK, 2022). (4) [CP Topici C [tP φPi T…[VP V…]]] A subtype within pragmatically licensed referential NSs are NSs licensed as a part of a “topic chain” (FRASCARELLI, 2007), that is, as successive occurrences of a previous overtly expressed Sentence Topic (REINHARt, 1981). this kind of topic chain is characteristic of MR, albeit in a very restrictive way, and licenses the presence of 3rd person NSs in matrix clauses, as we will see in Section 3.1. (b) The second mechanism available in PNS languages involves embedded clauses: an embedded NS can enter an anaphoric relation with respect to a c-commanding antecedent in the matrix clause, and thus receive a referential interpretation. This phenomenon is also known as finite control; see, i.a., Landau (2004) for Hebrew; Rodrigues (2004), Boeckx; Hornstein; Nunes (2010) for BP; Livitz (2014); tsedryk (2015) for MR. The embedded NSs in these contexts must be c-commanded by a coreferent antecedent in strict locality. 1902 Rev. Estud. Ling., Belo Horizonte, v. 30, n. 4, p. 1896-1935, 2022. (5) [DPi … [CP Opi C [tP φPi T…[VP V…]]]] In Section 4, I will revisit the structures introduced in this section, according to the specific languages and stages analyzed. 3 A compared characterization of NSs in EP-BP and OR-MR In this section, I will revise the “bundle of properties” that often go hand-in-hand with pro-drop, and which tend to cluster in a similar way in PNS languages (BIBERAuER, 2008). I will check these properties in BP and MR, and compare them with their CNS “ancestors”, EP, and OR, respectively (cf. fn. 1). the properties reviewed are the following: (i) pragmatically licensed referential NSs; (ii) finite referential embedded NSs; (iii) NSs with arbitrary and generic readings; (iv) null objects; (v) clitic pronouns; (vi) verbal inflection; and (vii) the role of infinitives. 3.1 Pragmatically licensed referential NSs MR licenses 1st and 2nd person NSs related to logophoric topical features, on a hearer-speaker discourse basis (6a), in a similar way to several Germanic languages, analyzed by e.g., Sigurðsson (2011). According to Duarte (1993), written BP allows this kind of drop in restricted cases, e.g., with negation or in a verbal locution, as in (6b), although some authors indicate that NSs in root sentences are not generally licensed in BP (MODESTO, 2000), or that they are on their way to being lost (DuARtE; SILVA, 2016): (6) a.Privet! Xorošo, čto (vy) prišli! hi well that (you) came.pl ‘Hi! So good that you came!’ (MR - MADARIAGA, 2022a, p. 79) b.(Eu) não posso mais ficar aqui a tarde toda. (I) not can.1sg more stay here the evening whole ‘I cannot stay here the whole evening any more’ (BP - DuARtE, 1993, p. 119) In MR, this kind of dropping is blocked by any fronted referential overt argumental NP introduced between the CP and the NS (LIVITZ, 2014; tSEDRyK, 2015). In example (7), the fronted object blocks referential subject drop, and the only way to interpret the NS is as a generic in an impersonal sentence (see Section 3.3 below): Rev. Estud. Ling., Belo Horizonte, v. 30, n. 4, p. 1896-1935, 2022. 1903 (7) Svetu pro nedavno videli na rynke. Sveta.acc recently saw.pl at market ‘Sveta was seen recently at the market.’ / ‘*we saw Sveta recently at the market.’ (MR - tSEDRyK, 2015, p. 349) Additionally, MR licenses 3rd person NSs related to Given or Familiar Topics or as successive occurrences of a Sentence Topic in a topic chain (LIVItZ, 2014; tSEDRyK, 2015), as in (8a). As in the previous case, any fronted referential NP can block subject drop (8b). (8) a. Onii idut na ozero. proi Nadejutsja tam vstretit’ Ivanaj. they.nom go.3pl to lake hope.3pl there see Ivan.acc *(Onj) obeščal im peredat’ ključi. he.nom promised.m.sg them pass keys ‘They are going to the lake. (They) hope to see Ivan there. He has promised them to pass the keys.’ (MR - tSEDRyK, 2022, p. 42) b. Jai tol’ko čto videl Svetuj. Mnei *(onaj) skazala, I.nom just saw Sveta.acc me.dat she.nom said.f.sg čto naš dom uže prodan. that our house already sold ‘I have just seen Sveta. She told me that our house had already been sold.’ (MR - tSEDRyK, 2015, p. 349) Notice that Sentence topics themselves cannot be dropped in MR; thus, in example (8a), the subject of the second sentence in this sequence is the first identical occurrence of the initial subject oni “they” in a topic chain and, therefore, can be dropped. However, the topic chain finishes here, because the third sentence introduces new propositional content; i.e., it shifts from “us going to the lake and doing something there” to “information about Ivan”. Thus, the new subject, Ivan, qualifies as a Sentence Topic and cannot be dropped. In example (8b), the presence of a fronted indirect object (mne) blocks subject drop, and the pronominal subject must be overtly realized. In BP, it seems that a subject can be dropped only in those cases in which it is topically very prominent, as in question-answer pairs (9a). Barbosa; Duarte; Kato (2005) also report examples of 3rd person referential NSs in BP tied to Topics; they state that it is required “that the antecedent is in an adjacent sentence”, as shown in (9b). The latter are very similar to those NSs inserted in topic chains in MR (8a). 1904 Rev. Estud. Ling., Belo Horizonte, v. 30, n. 4, p. 1896-1935, 2022. (9) a. – E o Joãoi? – proi Viajou. and the Joao? travelled.3sg ‘– what about Joao? – He travelled.’ (BP - DuARtE; SILVA, 2016, p. 239) b. O homemi finge que é um certo tipo de homem the man pretends that is a certain type of man para escrever. Ou seja, proi trai o homem. to write or be betrays the man ‘A man pretends to be a certain type of man in order to write. In other words, he betrays himself.’ (BP – BARBOSA; DuARtE; KAtO, 2005, p. 26) As for OR and EP / early BP, as well-behaved CNS languages, they show no locality effects in licensing referential NSs. Example (10a) shows that, in OR, a NS can take its reference not from the closest subjectual antecedent, i.e., the patriarch, but from a previous Sentence topic, Olga. In MR, there would need to be an overt subject “she” in the place of the NS; otherwise, the antecedent of the NS would have to be interpreted as the patriarch, the last overt subject previously mentioned. Similarly, in EP an overt pronoun tends to be avoided unless its reference is impaired (BARBOSA; DuARtE; KAtO, 2005), but the lack of proximity, or the syntactic position of the antecedent, as well as the presence of intervening elements, as in (10b), are not a problem for a NS to be licensed. According to Duarte (1993), in early BP texts, the rate of NSs was similar to that of EP, so we can assume that the situation was similar in both languages. (10) a. I bl(ago)s(lo)vi jui patrearxъj. I proi ide blessed.3sg her.cl.acc.f patriarch.nom and went.3sg s miromъ vъ svoju zemlju. (OR - Laurentian Chronicle 18) with peace to own land ‘And the patriarch blessed her (=Olga). And (she) went to her land in peace.’ b. As coisas não têm corrido tão bem como [as Nações unidas]i the things not have run so well as the nations united queriam, (...). A violência não pode ser dinamizadora wanted.3pl the violence not can be propeller da mudança. Mas muito rapidamente proi concluíram que of the change but very quickly concluded.3pl that todas as condições estavam reunidas. all the conditions were gathered ‘Things are not going as well as the United Nations expected (…). Violence cannot be the propeller of change. But very quickly they (=the UN) concluded that all the conditions were gathered.’ (EP – BARBOSA; DuARtE; KAtO, 2005, p. 27.) Rev. Estud. Ling., Belo Horizonte, v. 30, n. 4, p. 1896-1935, 2022. 1911 example of the co-occurrence of an aorist (reče ‘he said’), and a perfect compound form (perekljukala esi ‘you deceived’): (22) I reč(e) c(ěza)rь: perekljukala mja esi, Olьga. and said.3sg tsar deceived.f.sg me.cl aux.2sg Olga ‘And the tsar said: “you tricked me, Olga.”’ (OR – Laurentian Chronicle 17v) 3.7 The role of infinitives Infinitives underwent a process of impoverishment in BP in parallel with that experienced by finite forms. EP displays four unambiguous forms out of five total forms: eu / ele/ela cantar ‘I / he/she to sing’, tu cantares ‘you (sg) sing’, nós cantarmos ‘we sing’, vós cantardes ‘you (pl) sing’, eles cantarem ‘they to sing’. In contrast, BP infinitives display no unambiguous forms: eu / ele/ela / a gente cantar ‘I / he/she / we to sing’, vocês / eles cantarem ‘you (pl) / they to sing’ (cf. i.a., DuARtE; GONÇALVES; SANTOS, 2012; MODEStO, 2018; NuNES, 2011; RAPOSO, 1987). the only inflectional form for the infinitival in BP is 2nd-3rd person plural -em, alternating with the zero ending (23), while other persons and numbers invariably show a zero ending. (23) Eles saíram depois de jantar / jantarem cedo. they left after of dine.inf / dine.inf.3pl early ‘They left after having dinner early.’ (BP – NUNES, 2011, p. 336) Neither OR or MR displayed inflected infinitives, but OR had the possibility of using overt dative subjects in embedded infinitive clauses. These subjects could be non-coreferent DPs or coreferent overt strong pronouns in strong or emphatic positions (24a), alternating with coreferent non-emphatic null pronouns (24b). This alternation was completely parallel to the interplay between the overt emphatic (25a) and null non-emphatic pronouns in finite clauses (25b) (cf. MADARIAGA, 2018). (24) a. [Ty so mnoju]i cělovalъ kr(e)stъ [xoditi namai you with me kissed.SG cross go.INF we.DAT po odinoj dumě oběma]. by one decision both.dat ‘You and me swore (lit. kissed the cross) that we both (i.e. not only me) would do it the same way.’ (OR – Laurentian Chronicle, 71R) b. Egdaž(e) trebuetь na voinu iti, siii xotjat(ь) if is needed to war go these want.3pl 1912 Rev. Estud. Ling., Belo Horizonte, v. 30, n. 4, p. 1896-1935, 2022. [proi počti c(ěza)rja vaše(go)] (. . .) da budutъ. honour.inf tsar your let be.pl ‘If you need to gather an army (lit. to go to war), and these (=the Russians) want to join your tsar (. . .), so be it.’ (OR – Laurentian Chronicle, 18R) (25) a.Vižь sego ty ježe esi xotělъ. look this you what aux.2.sg wanted.m.sg ‘See, this is what you (i.e., but not me) wanted.’ (OR – Laurentian Chronicle, 23R, MADARIAGA, 2018, p. 181) b. Počto proi ideši opjatь, proi poimalъ esi what for go.2.sg again took.m.sg aux.2.sg vsju danь. whole tax ‘why did you come back? you collected the tribute already’. (OR – Laurentian Chronicle, 14R, MADARIAGA, 2018, p. 181) Overt infinitive subjects alternating in this way are not found in EP or early BP. As we will see in Section 4.1, this property of OR will be significant for the rise of embedded control in MR. 3.8 Interim summary In table 2, I offer a revised summary of the similarities and differences between the four relevant languages with respect to the features usually ascribed to the pro-drop parameter: table 2 – clustering properties ascribed to the pro-drop parameter in OR, MR, EP, and BP OR MR EP BP 1. Referential NSs free restricted free restricted 2. Finite control no yes no yes 3. Generic 3sg NSs no no no yes 4. Null objects no yes restricted yes 5. Clitic objects yes no yes restricted 6. Finite verbal person inflection yes yes (only non-past tenses) yes very defective 7. Inflected infinitivals no no yes very defective In view of these correlations, we can affirm that the change in Russian pro-drop affected defective referential pronouns (referential null subjects, including finite control, and clitics / null objects), but it did not have a strong impact on generic/arbitrary NSs or verbal inflection. Rev. Estud. Ling., Belo Horizonte, v. 30, n. 4, p. 1896-1935, 2022. 1913 In BP, all defective subjects (referential or generic) changed, together with a clear impoverishment in verbal inflection (finite and non-finite), and perhaps minor consequences concerning defective objects (null or clitic). In the following section, I offer a unified account for these facts. 4 An explanation for the change in pro-drop in BP and MR In this section, I will develop the idea that variation among PNS languages is the effect of historical change, whose contingent properties can partially shape grammars.5 The fact that pro-drop is triggered by the confabulation of a few recurrent factors, which can be diachronically modified in different ways, gives us the slight differences between BP and MR with respect to their pro-drop properties, rendering thus a satisfactory explanation for the parametrical “mismatches” we observe in these PNS languages. I will show: (i) how the final setting of the PNS parametric stage was essentially the same in both BP and MR; and (ii) in which way the triggers of the change were different in the two languages, and why the clustering ways of the properties reviewed above also differed at some points. 4.1 Structures and change in NS licensing in Russian the most striking fact about referential NSs in Russian (27a-b), contrasting with Hebrew (26a-b) or Chamorro, is that their conditions of licensing, described in Section 3.1, are completely equal regardless of the richness or the poverty of personal inflection on the verbal form. that is, whether the verbal form displays morphologically overt personal features (in non-past tenses, endowed with overt personal morphology, 27b) or not (with past impoverished l-forms, lacking personal morphology, 27a), the 5 I assume a Lightfootian (LIGHTFOOT, 1999, and subsequent work) approach to change. In neutral conditions, any modification of the Primary Linguistic Data (PLD) can lead learners to diachronic change, i.e., parse a different grammar with respect to previous generations. However, change is contingent on the unpredictability of the external factors conditioning it. The most frequent triggers modifying the PDL that learners receive are the following: (i) changes that have taken place previously in the language and can condition the “clustering” effects, like the ones we are considering in this paper (traditionally termed “chains of changes” in historical linguistics); (ii) phonological erosion or attrition (which can have an impact on morphological material); (iii) language contact and other sociolinguistic factors. 1914 Rev. Estud. Ling., Belo Horizonte, v. 30, n. 4, p. 1896-1935, 2022. conditions for licensing referential NSs apply equally; in example (27), logophoric features license dropping of any of the subjects. In Hebrew, however, rich agreement licenses subject dropping (26a), while defective non-personal agreement does not (26b): (26) a. Etmol šama’t harca’a. yesterday hear.past.2sg.f lecture ‘yesterday you heard a lecture.’ (Hebrew – DORON, 1988, p. 205) b. *Axšav šoma’at harca’a. now hear.pres.sg.f lecture Intended: ‘you are hearing a lecture now.’ (Hebrew – DORON, 1988, p. 207) (MR) (27) a.(Ty) v magazin xodil, ili doma sidel ves’ den’ you to shop go.past.m.sg or home sit.past.m.sg all day pered televizorom? before TV ‘Have you gone to the shop or were you sitting in front of the TV all day long?’ b. (Ty) v magazin pojdëš’ ili doma budeš’ sidet’ you to shop go.fut.2sg or home fut.2sg sit pered televizorom? before TV ‘Will you go shopping or are you going to be sitting in front of the TV?’ The diachronic process that gave rise to the PNS character of Russian evidences the fact that the role of verbal inflection in the change was marginal (cf. MEyER, 2011; JuNG, 2018, contra MÜLLER, 2006); changes in verbal inflection did not constitute the initial trigger for the change and, in fact, verbal inflection only changed in past tenses (as a side-effect of a previous change, as we will see), whereas present and future tenses preserved their inflection paradigms completely untouched. Let us briefly review the process of loss of the referential NSs in Russian step-by-step. the following description is based on the works by Borkovskij (1978), Borkovskij; Kuznecov (1965), Ivanov (1990), Jung; Migdalski (2015), Kibrik (2013), Meyer (2011), Migdalski (2013), and Zaliznjak (2008). Rev. Estud. Ling., Belo Horizonte, v. 30, n. 4, p. 1896-1935, 2022. 1915 (1) Old Russian was a well-behaved CNS language (cf. Section 3). Referential subjects were dropped in pragmatically “neutral” conditions, regardless of the type of topic characterizing the NS, as shown in example (10a, 25b) above. Referential subjects had to be overtly realized as pronouns only in emphatic positions (foci or contrastive topics, ex. 25a). This was the situation until approximately the 14-15th century. the verbal system was inflectionally rich, there existed 4 past forms, a present, and several forms to express the future. However, in OR, some tense paradigms started to fall into disuse: in early Slavic there were two synthetic past forms (the imperfect and the aorist) and two analytic forms (perfect and pluperfect). Analytic verbs were formed by an auxiliary conveying person and number morphemes, and a participle (an l-form) conveying the lexical verbal content plus gender and number morphemes (examples 21a and 22 above). The auxiliaries corresponding to the 3rd person singular and plural had already been lost by early OR, and the synthetic past forms were in decay, too: the imperfect was almost inexistent, and the aorist was archaic, restricted to literary language. The fall into disuse of most tense distinctions led to the eventual rearrangement of the Russian verbal system in favor of aspectual distinctions (the MR pattern) rather than tense distinctions. Deficient pronouns other than subjects, as well as auxiliary verbs, were formally clitics. Auxiliary verbal forms, just like other clitics, were raised to a high position in the sentence, as in examples (21a) and (25a) above. Synthetic verbal forms were also raised higher (compare the OR example (28) and the later Middle Russian example (29b) below). this high verbal position is common to other CNS languages and is usually considered to reflect the existence of V-to-t movement in these languages. (28) i pro poklanęju ti sę. and bow.1sg you.cl.dt refl.cl.acc ‘I greet (lit. bow) you.’ (OR – Birch bark letter 798, 12th c., JuNG, 2018, p. 105) Pronominal clitics in OR raised to be adjacent to the raised verb or auxiliary (MIGDALSKy, 2013; JuNG, 2018). they were used in the same pragmatic situations as NSs, that is, in non-emphatic non-stressed positions (28), while strong pronouns were used in contrastive or focused positions. 1916 Rev. Estud. Ling., Belo Horizonte, v. 30, n. 4, p. 1896-1935, 2022. (2) Early Middle Russian: by the 14-15th century, verbal auxiliaries started to surface in a lower position (29a), and synthetic verbs also remained low in the structure (29b); compare the position of the synthetic verbal form in the OR example (28) and the Middle Russian one (29b). In other words, long verb movement got lost. (29) a. estъ u menei edinъ s(y)nъj doma menšii, a proi sь četyrmi is at me one son at home young and with four esmь vyšelъ, a onъj doma. aux.1sg left.m.sg and that at.home ‘I have a little son, and I came here with my other four children, and the young one stayed at home.’ (Middle Russian – Hypathian Chronicle 46, 14th c.) b. jęza tobe koloneju-sę. I.nom you.strong.dt bow.1sg-refl ‘I greet you.’ (Middle Russian – Birch bark letter 501, 14th c., JuNG, 2018, p. 105) Very soon, between the 15th and 16th centuries, verbal auxiliaries (of 1st and 2nd person; remember that 3rd person auxiliary had been lost long ago) were lost in the language, after a short period of lowering. At the same time, pronominal clitics were also being lost, and being replaced with strong pronominal forms. Auxiliary clitics were not eroded phonologically or progressively worn out; the whole forms, all of them disyllabic (sg. esmь, esi, pl. esme, este), were lost altogether (30). Almost simultaneously, overt pronouns of 1st and 2nd person were realized in neutral positions (non-emphatic, non-stressed), as in (30) and (29b), in the places where formerly only NSs were used (28). (30) A az stal v dolu s polkom, and I.nom stood.l-form.m.sg in valley with army, a Vasilьju prikazal… and Vasili.dt ordered.l-form.m.sg ‘And I stood in the valley with the army and I ordered Vasili…’ (Middle Russian – Pervoe pis’mo Vasilija Grjaznogo Ivanu IV Groznomu 1576, 20) This rendered a new pattern with “impoverished” personal morphology, but only in past tense, which happened to be the one formerly including a clitic auxiliary (which, in turn, happened to be the part of the verbal complex bearing personal morphology). At the same time, present and future personal morphology remained intact but, Rev. Estud. Ling., Belo Horizonte, v. 30, n. 4, p. 1896-1935, 2022. 1917 nonetheless, showed similar rates of overt non-emphatic pronominal subjects as past impoverished forms (MEyER, 2011). In other words, overt non-emphatic pronouns were realized in the place of NSs at a fixed rate regardless of the “richness or poverty” of agreement in the specific verbal form in the sentence. (3) Late Middle Russian: by the 16-17th centuries, when the modern system of 1st and 2nd pro-drop was already well-established, 3rd person sentences experienced the same shift as 1st and 2nd person. 3rd person verbal auxiliaries in perfect > past tense had already been lost in early OR (approx. 5-6 centuries before), but no personal pronoun existed in the language for the 3rd person. In stressed or emphatic positions, when the pronominal subject had to be overtly realized, various demonstrative pronouns were realized instead. One of these demonstratives, onъ ‘that’, was reanalyzed in Late Middle Russian as the personal pronoun of 3rd person and immediately spread to non-emphatic and nonstressed positions (31), following the pattern of the 1st and 2nd person pronouns. (31) Prosi y nego na brašna deneg, i on tebe dast sto rublej. ask from him for food money and he you will give hundred rubles ‘Ask him for money for food, and he will give you a hundred rubles.’ (Middle Russian – Story about Karp Sutulov, 17th century) As for pronominal clitics, 1st and 2nd person object clitics were lost by the 15-16th century, while 3rd person clitics were lost approximately a century later. The delay in the loss of 3rd person clitics as compared to 1st and 2nd person parallels the development of 3rd person vs. 1st / 2nd person overt pronominal pronouns in the place of the old NSs. Null objects are found in texts roughly by the 16th century, and became more common later on: (32) A ženixi po nevestuj ne ezdit, a privezet proj družka da svaxa. and groom for bride not goes but carries best man and matchmaker ‘The groom does not go for the bride; the best man and the matchmaker bring (her).’ (Middle Russian – Putešestvija russkix poslov, BORKOVSKIJ, 1978, p. 314) In Table 3, I summarize the development of the change in the relevant pro-drop features in Russian: 1918 Rev. Estud. Ling., Belo Horizonte, v. 30, n. 4, p. 1896-1935, 2022. Table 3 – diachronic succession of changes in the pro-drop system between OR and MR When Changes 11-12th c Loss of 3rd person auxiliary Loss of past tenses in favor of aspectual distinctions 15th c Loss of long verb movement = lowering of auxiliaries and synthetic verbs 15-16th century Loss of 1st/2nd auxiliaries Extension of 1st/2nd overt pron. subjects Loss of 1st/2nd pron. object clitics 16-17th century Extension of onъ as 3rd p. overt pronoun Loss of 3rd person object clitics Extension of null objects Now let us interpret the change in formal terms (cf. MADARIAGA, 2022a; 2022b). As I explained in Section 2, in OR, the minimal φP subject, located at [Spec,T], automatically received a referential interpretation from a higher topic by D-matching. The referential index was copied by the unvalued D-feature of T, and, then, through Agree, by the NS, which at the same time matched its unvalued φ-features. the D-feature of t was “spelled out” in the form of rich verbal morphology. Overt V-to-T movement conveyed the presence of rich agreement morphology high in the sentence, which constituted a clear cue for learners to establish the presence of a [+D] feature on T. (33) [CP Topici C [TP φPi TD:_+V [VP V…]]] The loss of 3rd person auxiliaries and the obsolescence of most past forms in early OR led to a reorganization of the tense – aspect system in Russian. Because of the loss of 3rd person verbal auxiliaries, analytic past forms in 3rd person could seem to speakers like low “synthetic” l-forms, marked for gender and number (not person), which did not raise to T. V started to be interpreted as undergoing short movement (to Asp or a similar intermediate node, as it does nowadays (cf. GRIBANOVA, 2013); confirmed by experimental work by Kallestinova; Slabakova (2008)), and the old tense-based system disappeared in favor of a new system based on aspect distinctions. As for non-past synthetic forms, they preserved person marking but, analogically to past l-forms, they stopped moving to T. 1st and 2nd person auxiliaries also displayed a low position at this time (ZALIZNJAK, 2008). In other words, V-to-t movement was lost. The loss of long verb movement modified the cue that learners needed to posit an unvalued D-feature on T, that is, the overt Rev. Estud. Ling., Belo Horizonte, v. 30, n. 4, p. 1896-1935, 2022. 1919 morphological realization of the operation of D-feature valuing, which was person agreement on T (cf. a similar idea in Jung, 2018, who proposes D-feature lowering rather D-feature loss). The loss of overt morphology at T prevented the acquisition of T as playing a role in D-feature transmission together with Agree. In addition, once V remained in AspP, V’s low position preempted the establishment of the direct syntactic relation between T and C the way it did before. The loss of the D-feature on T was completed between the 15th and 16th centuries, with the total loss of personal auxiliaries in the language. After T’s D-linking ability was lost, pronominal subjects had to be overtly realized in order to be interpreted as referential. However, learners of Middle Russian kept on receiving referential subject gaps in their linguistic input in the form of NSs generated by older generations of speakers, and subject gaps in control infinitive clauses (PRO). As a common output in historical processes, residual structures and elements after diachronic change can die out or be “recycled” (reanalyzed) with a new value. the old NSs in Russian experienced the second path. Speakers of Middle Russian, in the absence of a D-linking t, had two ways to acquire the finite subject gaps they received: (i) as generic or arbitrary NSs (they retained 3rd person plural, as in OR), or (ii) as referential NSs when they were able to find a proper mechanism of index transmission, whether a prominent or logophoric null topic at CP, or a higher c-commanding referential antecedent in the case of embedded NSs. As for other non-subjectual deficient pronominals, namely, pronominal clitics, the change reviewed here also had a direct impact on them. I will assume the classic proposal by Cardinaletti; Starke (1999); namely that prosodic, phonological, and morphological deficiency of clitics correlates with less syntactic structure, as compared to full pronouns. Thus, a clitic is a minimal noun projection, i.e., just an agreement projection. In our terms, clitics would have the same basic structure as NSs, i.e., minimal φPs with unvalued interpretable φ-features (φP[iφ:_]). Cf., i.a., BLEAM, 1999; DuARtE; MAtOS; GONÇALVES, 2005; ROBERtS, 2010; BOŠKOVIĆ, 2016, for a similar idea.6 6 OR deficient pronouns qualify as φP clitics, rather than mere agreement markers (cf. contrastive properties of the two types of clitics in Bleam, 1999; Ormazabal; Romero, 2013; and references therein). Accusative clitics in Slavic, including its earlier periods, 1920 Rev. Estud. Ling., Belo Horizonte, v. 30, n. 4, p. 1896-1935, 2022. As in the case of NSs, OR clitic φPs are not D-complete elements, so they must have their reference defined in the course of the derivation. As discussed before, the reference of a minimal φP is determined by some (null or overt) Topic at CP. Getting a reference from this high Topic conveys also φ-feature valuation on the clitic, as in the case of NSs. In the case of CNS languages, licensing minimal φPs (NSs and clitics) is tied to the presence of an unvalued D-feature on t (HOLMBERG; NAYUDU; SHEEHAN, 2009), which copies the referential index of the higher Topic, and transmits it downwards during the Agree operation of φ-feature matching by the φP. In the case of subjects, as a result of Agree, the valued features of T are “spelled out” in the shape of rich verbal morphology, and NSs can be realized as null. However, objectual φPs can receive their reference, but they lack the Agree relation with t described above for subjects, so they must be overtly realized, albeit in a prosodically deficient way. In the specific case of OR and other Slavic languages, following, i.a., Progovac (1999); Migdalski (2016); Bošković (2016), clitic licensing is parasitic on verb movement, like pro subjects in CNS languages. Being prosodically deficient, in the specific case of OR, objectual φPs had the requirement to be adjacent to the verb (JuNG; MIGDALSKI, 2015). thus, when the verb raised in OR, the φP moved together with the verb up to their final common landing site, in which V attached to t. In this pattern, both elements moved “for a reason”; the V to get its inflectional suffix, and the clitic to get prosodic support and referential interpretation. (34) [CP Topic1 Topic2 … C [TP φP1 (=NS) t-v-V φP2 (=clitic) [vP φP1 v+V φP2 [VP V φP2]]]] After V-to-T movement was lost, clitics could not raise, and had to remain low in the structure, together with their prosodical host, the verb. The clitics’ new low position, however, led to their unavoidable loss, as they were too far from C, and could not get a suitable reference. Therefore, very soon clitics were completely replaced by either (i) full, phonologically independent, and syntactically complete pronouns, or (ii) null objects, which, following the new reanalyzed pattern of the other null φPs in the language, the NSs, could raise as much as they needed cannot double DPs or override Principle B, in contrast to clitics in other languages, which are better analyzed as agreement elements. Rev. Estud. Ling., Belo Horizonte, v. 30, n. 4, p. 1896-1935, 2022. 1927 intermediate position, as in Russian), as a side-effect of the loss of rich person agreement. It seems that the change in pro-drop in BP and in Russian represented two sides of the same coin: in Russian, the loss of V-to-t movement modified the cue that learners needed to posit an unvalued D-feature on T, the overt morphological realization of the operation of D-feature valuing, i.e., person agreement overtly realized in a high position (at T). Thus, T was considered not to take part in D-feature valuing any more, and the D-feature of T was lost. In BP, on the other hand, the overt morphological realization of D-feature valuing, person agreement, was lost altogether, and this automatically led to the loss of the D-feature on T. At the same time, the loss of person agreement had the side-effect that V-to-t movement could be perceived by BP learners as unmotivated and decayed as a result. Clitic objects did not change in BP as much as in Russian, in which clitics were totally replaced by strong pronouns or newly created null objects. In BP, 1st and 2nd person clitics were preserved, but the low position of the verb brought with it a change in the cliticization direction of clitics. Duarte; Matos; Gonçalves (2005) claim that variation in clitic order, and its different patterns in EP and BP depend on the ability of t and Asp to attract V versus T’s capacity to check uninterpretable features through Agree without attracting V. In EP, clitics are always enclitic, whether the preceding word is a verb or not (NUNES, 2011); when the verb is attracted to T, the clitic “remains” to its right, as in (19b) above, and in case there is some element at C (a wh-word, negation, etc.), leftward cliticization takes place with respect to the C-element (21b). On the other hand, in BP the verb remains low and the clitic is spelled-out “to its left”; thus, learners interpret this pattern as proclitic cliticization. In this sense, the change from enclisis to proclisis in the 1st and 2nd person clitics in BP could be a further consequence of the change in the position of V after the loss of rich person agreement. With regard to the loss of 3rd person clitics, Nunes (1993) argues that it was due to the change of directionality in cliticization in BP. The new system of proclisis made impossible for learners to acquire 3rd person clitics, whose syllabic onset was incompatible with rightward cliticization.8 The loss of 3rd person clitics led to the emergence of weak 8 Nunes (1993) notes that 3rd person clitics in Portuguese (o, a, os, as) lost the onset of their syllable (the initial l-, present in other Romance languages, like the Spanish lo, la, los, las). He 1928 Rev. Estud. Ling., Belo Horizonte, v. 30, n. 4, p. 1896-1935, 2022. pronouns in object position, as well as the expansion of the already existing null object constructions. As was the case in Russian, the new ways of licensing referential NSs in pragmatically motivated specific situations in BP paved the way to reanalyze the old null objects, traces of wh-movement in EP (RAPOSO, 1986), as real null pronouns (GALVES, 1989, i.a.) by analogy with the new NSs. Thus, null objects in BP became more unrestricted from the grammatical point of view than in EP, as they only needed to be pragmatically motivated. In that sense, the final result of the change in MR and BP, being initially quite different, resulted in a similar output. Finally, arbitrary sentences in BP retained the old patterns (3rd person plural agreement and the se-construction), and created a new pattern with 3rd person singular agreement, which arose after the loss of the se clitic element (NUNES, 1991), probably in relation to the loss of 3rd person clitics in general in the language. this process differs from Russian, because in OR, unlike EP, impersonal sentences were never formed with a sja clitic (equivalent to Romance se clitic), and thus the corresponding arbitrary construction in MR did not have a suitable ancestor with 3rd person singular agreement, on the basis of which learners could build the construction that arose in BP. 5 Conclusion In this paper, I have shown that the contingent nature of change determines the fact that one and the same parametric change can be biased by slightly different historical conditions, which can render slightly different results. However, the final parametric setting will be very similar, as proven by the detailed examination of the change in pro-drop, experienced by MR and BP independently from each other. These two languages developed from a CNS stage into a PNS stage for different reasons and according to different pathways of change, but ended up having basically the same properties. From a parametric argues that the onset of the syllable of 3rd person clitics is licensed via phonological cliticization to the preceding word as happens in EP, in which clitics are always enclitic to whatever word they need to be. However, in BP cliticization is always rightward, which prevented the onset of the syllable of 3rd person clitics to be phonologically licensed, and led to their disappearance. 1st and 2nd person clitics (me, te, nos, vos) did not lose their syllabic onset, so they were preserved in BP, albeit in the new position. Rev. Estud. Ling., Belo Horizonte, v. 30, n. 4, p. 1896-1935, 2022. 1929 point of view, both languages lost the D-feature on T, associated with a series of common properties (short verb movement, reanalysis of NSs as pragmatically licensed, automatic interpretation of a NS as arbitrary or generic, finite control, expansion of null objects, weakening of clitics, reanalysis of embedded NSs), even if the exact pathways of change differed between the two languages. the small differences in the final results of the change have been attributed to different initial conditions in the CNS stage of pro-drop; namely, the existence or not of impersonal se constructions, the role of the impoverishment of agreement, the differences in non-finite embedded clauses, and the initial trigger of the change (reorganization of the tense–aspect system vs. rearrangement of the pronominal system). Acknowledgements This research has been supported by the PGC2018-096870-B-100 and PGC2018-098995-B-I00 projects, funded by the Spanish Ministry of Science, Innovation, and universities, the Spanish Research Agency, and the European Regional Development Fund (FEDER), and the It134419 research group, funded by the Basque Government. The author is especially grateful to Maia Duguine for advice and help with this paper, and to two anonymous reviewers at the Revista de Estudos da Linguagem for their comments and suggestions. I also want to thank Xavier Cruz and Pavel Graschenkov for help with European Portuguese and Russian data, respectively. References BARBOSA, P; DuARtE, M. E. L.; KAtO, M. A. Null Subjects in European and Brazilian Portuguese. Journal of Portuguese Linguistics, Lisboa v. 4, n. 2, p. 11-52, 2005. DOI:10.5334/jpl.158 BIBERAuER, t. Introduction. In: BIBERAuER, t. (ed.). The limits of syntactic variation. John Benjamins, 2008. p. 1-72. BIBERAuER, t.; ROBERtS, I.; HOLMBERG, A.; SHEEHAN, M. Introduction: parameters in minimalist theory. In: BIBERAuER, t.; ROBERtS, I.; HOLMBERG, A.; SHEEHAN, M. (eds.). Parametric Variation: Null Subjects in Minimalist Theory. Cambridge: Cambridge University Press, 2010. p. 1-57. 1930 Rev. Estud. Ling., Belo Horizonte, v. 30, n. 4, p. 1896-1935, 2022. BLEAM, T. Leísta Spanish and the syntax of clitic doubling. PhD Dissertation, University of Delaware, 1999. BOECKX, C.; HORNStEIN, N.; NuNES, J. Control as Movement. Cambridge: Cambridge University Press, 2010. BORKOVSKIJ, V. I. Istoričeskaja grammatika russkogo jazyka: sintaksis - prostoe predloženie. Moskva: Nauka, 1978. BORKOVSKIJ, V. I.; KuZNECOV, P. S. Istoričeskaja grammatika russkogo jazyka. Moskva: AN SSSR, 1965. BOŠKOVIĆ, Ž. On second position clitics crosslinguistically. In: MARuŠIČ, F.; ŽAuCER, R. (eds.). Formal studies in Slovenian syntax: In honor of Janez Orešnik. Amsterdam - Philadelphia: John Benjamins, 2016. p. 23-53. CAMACHO, J.; ELíAS-ULLOA, J. Null subject systems in Shipibo switch-reference. In: CAMACHO, J.; GutIÉRREZ-BRAVO, R.; SÁNCHEZ, L. (eds.). Information structure in languages of the Americas. Berlin: Mouton de Gruyter, 2010. p. 65-85. CARDINALEttI, A.; StARKE, M. the typology of structural deficiency: A case study of the three classes of pronouns. In: VAN RIEMSDIJK, H. (ed.). Clitics in the languages of Europe. Berlin: Mouton de Gruyter, 1999. p. 185-234. CORRÊA, V. O Objeto Direto Nulo no Português do Brasil. MA thesis, Universidade Estadual de Campinas, 1991. CyRINO, S. M. L. Observações sobre a mudança diacrônica no português do Brasil: Objeto nulo e clíticos. In: ROBERtS, I; KAtO, M. A. (eds.). Português Brasileiro: Uma Viagem Diacrônica, Campinas: Editora da UNICAMP, 1993. p. 163-184. CyRINO, S.; ORDOñEZ, F. Null objects in Brazilian Portuguese and DOM in Spanish: similarities and diferences. In: PARODI, t. (ed.). Proceedings of the VIII Nereus International Workshop: Referential Properties of the Romance DP in the Context of Multilingualism. Kontanz: Fachbereich Sprachwissenschaft, universität Konstanz, 2018. p. 103-120. DuARtE, M. E. L. Do pronome nulo ao pronome pleno: a trajetória do sujeito no português do Brasil. In: ROBERtS, I; KAtO, M. A. (eds.). Português Brasileiro: Uma Viagem Diacrônica, Campinas: Editora da UNICAMP, 1993. p. 107-128. DuARtE, M. E. L. the loss of the ‘avoid pronoun’ principle in Brazilian Portuguese. In: KAtO, M. A.; NEGRÃO, E. V. (eds.). Brazilian Rev. Estud. Ling., Belo Horizonte, v. 30, n. 4, p. 1896-1935, 2022. 1931 Portuguese and the Null Subject Parameter, Madrid & Frankfurt am Main: Iberoamericana & Vervuert, 2000. p. 17-36. DuARtE, M. E. L.; SILVA, M. C. F. the Null Subject Parameter and the Structure of the Sentence in European and Brazilian Portuguese. In: WETZELS, W. L.; COSTA, J.; MENUZZI, S. (eds.). The Handbook of Portuguese Linguistics, Malden, MA: Wiley-Blackwell, 2016. p. 234253. DuARtE, I.; GONÇALVES, A.; SANTOS, A. L. Infinitivo flexionado, independência temporal e controlo. In: COSTA, A.; FLORES, C.; ALEXANDRE, N. (eds.). Textos Seleccionados do XXVII Encontro Nacional da Associação Portuguesa de Linguística, Lisboa: APL, 2012. p. 217-234. DuARtE, I.; MAtOS, G.; GONÇALVES, A. Pronominal Clitics in European and Brazilian Portuguese. Journal of Portuguese Linguistics Lisboa, v. 4, p. 113-141, 2005. DOI: 10.5334/jpl.161 FRASCARELLI, M. Subjects, topics and the interpretation of referential pro. Natural Language and Linguistic Theory, Dordrecht, v. 25, p. 691734, 2007. DOI: 10.1007/s11049-007-9025-x GALVES, C. O objeto nulo no português brasileiro: Percurso de uma pesquisa. Cadernos de Estudos Lingüísticos, Campinas, v. 17, p. 65-90, 1989. DOI: XXXXX GALVES, C. O enfraquecimento da concordância no português brasileiro. In: ROBERtS, I; KAtO, M. A. (eds.). Português Brasileiro: Uma Viagem Diacrônica. Campinas: Editora da UNICAMP, 1993. p. 387-403. GALVES, C. V-movement, levels of representation and the Structure of S. Letras de Hoje, Porto Alegre, v. 29, n. 2, p. 35-58, 1994. DOI: XXXXX GRIBANOVA, V. Verb-stranding verb phrase ellipsis and the structure of the Russian verbal complex. Natural Language and Linguistic Theory, Dordrecht, v. 31, n. 1, p. 91-136, 2013. DOI: 10.1007/s11049-012-9183-3 HOLMBERG, A. Is there a Little Pro? Evidence from Finnish. Linguistic Inquiry, Cambridge, MA, v. 36, p. 533-564, 2005. DOI: 10.1162/002438905774464322 HOLMBERG, A. Null Subject Parameters. In: BIBERAuER, t; HOLMBERG, A.; ROBERtS, I.; SHEEHAN, M. (eds.). Parametric Variation: Null Subjects in Minimalist Theory. Cambridge: Cambridge University Press, 2010. p. 88-124. HOLMBERG, A.; NAyuDu, A.; SHEEHAN, M. three partial nullsubject languages: a comparison of Brazilian Portuguese, Finnish and Marathi. Studia Linguistica, Hoboken, New Jersey, v. 63, p. 59-97, 2009. DOI: 10.1111/j.1467-9582.2008.01154.x Rev. Estud. Ling., Belo Horizonte, v. 30, n. 4, p. 1896-1935, 2022. 1932 HOLMBERG, A.; ROBERtS. I. Parameters and the three factors of language design. In: PICALLO, C. (ed.). Linguistic Variation in the Minimalist Framework. Oxford: Oxford University Press, 2014. p. 61-81. IVANOV, V. V. Istoričeskaja grammatika russkogo jazyka. Moskva: Prosveščenie, 1990. JuNG, H. Null subjects and person in Old North Russian. In: GRKOVICMAJOR, J.; HANSEN, B.; SONNENHAuSER, B. (eds.). Diachronic Slavonic syntax. Berlin: De Gruyter, 2018. p. 95-123. JuNG, H.; MIGDALSKI, K. On the degrammaticalization of pronominal clitics in Slavic. In: SZAJBEL-KECK, M.; BuRNS, R.; KAVItSKAyA, D. (eds.). Formal approaches to Slavic linguistics: the first Berkeley meeting 2014. Ann Arbor: Michigan Slavic Publications, 2015. p. 143162. KALLEStINOVA, E.; SLABAKOVA, R. Does the verb move in Russian?. In: ANtONENKO, A.; BAILyN, J.; BEtHIN, C. (eds.). Formal approaches to Slavic linguistics 16: The Stony Brook meeting 2007. Ann Arbor: Michigan Slavic Publications, 2008. p. 199-214. KAtO, M. A. the distribution of pronouns and null elements in object position in Brazilian Portuguese. In: ASHBy, w.; PERISSINOttO, M. M. G.; RAPOSO, E. P. (eds.). Linguistic Perspectives on the Romance Languages. Amsterdam/Philadelphia: John Benjamins., 1993. p. 225-235. KIBRIK, A. A. Peculiarities and origins of the Russian referential system. In: BAKKER, D.; HASPELMAtH, M. (ed.). Languages across boundaries: studies in memory of Anna Siewierska. Berlin : De Gruyter, 2013. p. 227-262. LANDAu, I. the scale of finiteness and the calculus of control. Natural Language and Linguistic Theory, Dordrecht, v. 22, p. 811-877, 2004. DOI: 10.1007/s11049-004-4265-5 LIGHTFOOT, D. The Development of Language: Acquisition, change and evolution. Oxford: Blackwell, 1999. LIVITZ, I. Deriving silence through internal reference: focus on pronouns. Ph.D. Dissertation, New York University, 2014. MADARIAGA, N. Diachronic change and the nature of pronominal null subjects: the case of Russian. In: CASALICCHIO, J.; COGNOLA, F. (eds.). Null Subjects in Generative Grammar. A synchronic and diachronic perspective. Oxford: Oxford University Press, 2018. p. 171-98. MADARIAGA, N. Referential null subjects in Russian: a synchronic and diachronic overview. In: DALMI, G.; tSEDRyK, E.; CEGŁOwSKI, P. Rev. Estud. Ling., Belo Horizonte, v. 30, n. 4, p. 1896-1935, 2022. 1933 (eds.). Null subjects in Slavic and Finno-Ugric: Licensing, structure, and typology. Berlin: Mouton de Gruyter, 2022a. p. 75-103. MADARIAGA, N. Deficient pronouns in the history of Russian: null subjects and object clitics. Ms. UPV/EHU, 2002b. McSHANE, M. A Theory of Ellipsis. Oxford: Oxford University Press, 2005. MEyER, R. The history of Null Subjects in East Slavonic. A corpus based diachronic investigation. Habilitation thesis, university of Regensburg, 2011. MIGDALSKI, K. Diachronic source of two cliticization patterns in Slavic. In: SALVESEN, C. M.; HELLAND, H. P. (eds.). Challenging Clitics. Amsterdam: John Benjamins, 2013. p. 135-158. MODESTO, M. Null Subjects without “rich” Agreement. In: KAtO, M. A.; NEGRÃO, E. V. (eds.). Brazilian Portuguese and the Null Subject Parameter, Madrid & Frankfurt am Main: Iberoamericana & Vervuert, 2000. p. 147-174. MODESTO, M. Null subjects in Brazilian Portuguese: a critique of two possible analyses. In: tORRES MORAIS, M. A.; ANDRADE, M. L. (eds.). História do Português Caipira. Campinas: editora da UNICAMP, 2009. p. 99-123. MODEStO, M. Inflected infinitives in Brazilian Portuguese and the theory of Control. In: SANTOS, A. L.; GONÇALVES, A. (eds.) Complement Clauses in Portuguese: Syntax and acquisition: Issues in Hispanic and Lusophone Linguistics, vol. 17. Amsterdam: John Benjamins, 2018. p. 59-100. MÜLLER, G. Pro-drop and impoverishment. In: BRANDt, P.; Fuß, E. (eds.). Form, Structure, and Grammar. A Festschrift presented to Günther Grewendorf. Berlin: Akademie Verlag, 2006. p. 93-115. NEwMEyER, F. Against a Parameter-setting approach to language variation. Linguistic Variation Yearbook, Amsterdam, v. 4, p. 181-234, 2005. DOI: 10.1075/livy.4.06new NUNES, J. Se apassivador e se indeterminador: o percurso diacrônico no português brasileiro. Cadernos de Estudos Lingüísticos, Campinas, v. 20, p. 33-57, 1991. NuNES, J. Direção de cliticização, objeto nulo e pronome tônico na posição de objeto em português brasileiro. In: ROBERtS, I; KAtO, M. A. (eds.). Português Brasileiro: Uma Viagem Diacrônica. Campinas: Editora da UNICAMP, 1993. p. 207-222. Rev. Estud. Ling., Belo Horizonte, v. 30, n. 4, p. 1896-1935, 2022. 1934 NUNES, J. A Note on Wh-islands and Finite Control in Brazilian Portuguese. Estudos da Língua, Vitória da Conquista, v. 8, n. 2, p. 79-103, 2010. DOI: 10.22481/el.v8i2.1129 NUNES, J. On the diachronic reanalysis of null subjects and null objects in Brazilian Portuguese: triggers and consequences. In: RINKE, E.; KuPISCH, t. (eds). The Development of Grammar: Language acquisition and diachronic change. John Benjamins, 2011. p. 331-354. ORMAZABAL, J.; ROMERO, J. Object clitics, agreement and dialectal variation. International journal of Latin and Romance linguistics, Berlin, v. 25, n. 2, p. 301-344, 2013. DOI: 10.1515/probus-2013-0012 PIRES, A. Verb movement and clitics: Variation and change in Portuguese. In: BAtLLORI, M.; HERNANZ, M-L; PICALLO, C; ROCA, F. (eds.) Grammaticalization and Parametric Variation. Oxford: Oxford University Press, 2005. p. 48-59. PROGOVAC, L. Eventive to and the placement of clitics in SerboCroatian. In: KENESEI, I. (ed.). Crossing boundaries: Advances in the theory of Central and Eastern European languages. Amsterdam: John Benjamins, 1999. p. 33-44. RAPOSO. E. P. On the null object in European Portuguese. In: JAEGGLI, O.; SILVA-CORVALAN, C. (eds.). Studies in Romance Linguistics. Dordrecht: Foris, 1986. p. 373-390. RAPOSO, E. P. Case theory and Infl-to-Comp: the Inflected Infinitive in European Portuguese. Linguistic Inquiry, Amsterdam, v. 18, p. 85-109, 1987. REINHARt, t. Pragmatics and linguistics: an analysis of sentence topics. Philosophica, Valparaiso, v. 27, n. 1, p. 53-94, 1981. DOI: 10.21825/ philosophica.82606 RIZZI, L. Issues in Italian Syntax. Dordrecht: Foris, 1982. ROBERtS, I. A deletion analysis of null subjects: French as a case study. Ms, University of Cambridge, 2010. RODRIGuES, C. Impoverished morphology and A-movement out of case domains. PhD Dissertation, University of Maryland, 2004. SHEEHAN, M. On the difference between exhaustive and partial control. In: CASALICCHIO, J.; COGNOLA, F. (eds.). Null Subjects in Generative Grammar: A synchronic and diachronic perspective. Oxford: Oxford University Press, 2018. p. 141-170. SIGuRÐSSON, H. A. Conditions on argument pro-drop. Linguistic Inquiry, Amsterdam, v. 42, n. 2, p. 267-304, 2011. DOI: 10.1162/ ling_a_00042 Rev. Estud. Ling., Belo Horizonte, v. 30, n. 4, p. 1896-1935, 2022. 1935 SILVA, M. C. F. Main and embedded null subjects in Brazilian Portuguese. In: KAtO, M. A.; NEGRÃO, E. V. (eds.). Brazilian Portuguese and the Null Subject Parameter. Madrid & Frankfurt am Main: Iberoamericana & Vervuert, 2000. p. 127-146. tARALLO, F. Fotografías Sociolingüísticas. Campinas: Pontes, 1989. tESCARI NEtO, A. On Verb Movement in Brazilian Portuguese: A Cartographic Study. PhD Dissertation, università Ca’ Foscari di Venezia, 2012. tSEDRyK, E. Deriving null pronouns: A unified analysis of subject drop in Russian. In: SZAJBEL-KECK, M.; BuRNS, R.; KAVItSKAyA, D. (eds.). Formal approaches to Slavic linguistics: the first Berkeley meeting 2014. Ann Arbor: Michigan Slavic Publications, 2015. p. 342361. tSEDRyK, E. Null subjects and their overt counterparts in East Slavic root clauses: referential and non-referential readings. In: DALMI, G.; tSEDRyK, E.; CEGŁOwSKI, P. (eds.). Null subjects in Slavic and Finno-Ugric: Licensing, structure, and typology. Berlin: Mouton de Gruyter, 2022. p. 35-74. uRIAGEREKA, J. Aspects of the syntax of clitic placement in western Romance. Linguistic inquiry, Amsterdam, v. 26, n. 1, p. 79-123, 1995. VAINIKKA, A.; LEVy, y. Empty subjects in Finnish and Hebrew. Natural Language and Linguistic Theory, Dordrecht, v. 17, p. 613-671, 1999. DOI: 10.1023/A:1006225032592 wALKDEN, G. Refining the “null argument cycle”: the place of partial null argument languages. In: 14TH INtERNAtIONAL DIACHRONIC GENERAtIVE SyNtAX CONFERENCE, Lisbon, 2012. Available at: <http://walkden.space/Walkden_2012_Lisbon.pdf >. Last accessed: 23 March 2022. ZALIZNJAK, A. A. Drevnerusskie ènklitiki. Moskva: Jazyki slavjanskoj kul’tury, 2008.