'Diachronic change and the nature of pronominal null subjects: the case of Russian'. Null Subjects...
Transcript of 'Diachronic change and the nature of pronominal null subjects: the case of Russian'. Null Subjects...
1
Chapter 7
Diachronic change and the nature of pronominal null subjects:
The case of Russian *
Nerea Madariaga
University of the Basque Country UPV/EHU
Abstract. This chapter focuses on the nature of null referential subjects (pro) through a case study in
Russian. The loss of pro-drop in Middle Russian implied that: (i) Null subjects in non-embedded
contexts became restricted to instances licensed by pragmatics. (ii) In embedded contexts, learners lost
the possibility of parsing pro in the embedded subject gap, and started to parse the alternative null
category available, PRO /or trace. Afterwards, silent embedded subjects (both finite and non-finite)
became licensed only by Obligatory Control. The unified way of licensing null subjects in embedded
contexts was determined by this diachronic process which confronted learners with two alternative
elements to be parsed in the relevant gap, and had to imply some lexical or featural content for the
referential pronoun (pro) as opposed to PRO /or trace, contradicting views like the rich agreement
hypothesis.
Keywords. history of Russian, pro-drop, null subjects, embedded clauses, Obligatory Control, pro, trace,
copy.
7.1. Introduction
* I would like to thank the audience of the workshop Understanding pro-drop: a synchronic and diachronic
perspective, as well as the editors and reviewers of this volumen, for their questions and remarks, which have
much contributed to improving this paper, and especially, Maia Duguine and Mara Frascarelli for additional
discussion after the meeting. I also want to thank Egor Tsedryk for his insightful discussion and help with the
Russian data, as well as my informants Yulia Adaskina, Sasha Arkhipov, Liosha Leontiev, Natalia Petrovskaya
and Masha Zalivanskaya. This work is part of the FFI2011-29218, FFI2014-53675-P and FFI2014-57260-P
research projects, funded by the Spanish Ministry of Science and Innovation, and has been supported by the
research group on linguistics UFI11/14 (funded by the UPV/EHU) and the research group on historical linguistics
IT 698-13 (funded by the Government of the Basque Country).
2
This chapter provides a formal analysis of the role that null subjects (NS) played in the process
of restructuring the system of Russian embedded (non-finite and finite) clauses. More
specifically, I will show that, being in principle independent phenomena, a detailed analysis of
the development of the Russian embedded clauses contributes in a significant way to the
understanding of the nature of pro itself.
The literature on the topic includes different insights regarding the properties of
pronominal NSs. Here, I will highlight a specific dichotomy which will be central in my
analysis, namely, the two different views on the nature of NSs (cf. Roberts and Holmberg 2010):
(A) NSs are real pronouns (of the special null category pro or deleted pronouns): Cardinaletti
(1997), Holmberg (2005), and Roberts (2010a), following an original idea by Rizzi (1986a),
with different nuances, assume that NSs are occurrences of phonologically unrealized or empty
pronouns (pro). Holmberg (2005) argues in favour of the hypothesis that pro has interpretable
features, occupies the Spec,TP position, and functions as an overt pronoun. More specifically,
Holmberg (2005) and Roberts (2010a) view pro as a deleted pronoun. In the case of deleted
pronouns, the fact that they are silent is thus a PF matter, while in case we assume a special null
category pro, it must already be null when it enters the derivation from the lexicon.
(B) NSs are not pronouns but something else (such as rich agreement or elliptical gaps, for
example): Different alternative explanations to the existence of a real silent pronoun have been
pursued within this view. Here, I will mention the most successful of them, namely the
hypothesis that, in the so-called consistent NS-languages, the subject can be directly expressed
by rich verbal agreement inflection (cf. Barbosa 1995; Alexiadou and Anagnostopoulou 1998,
based on Borer 1986).
In this chapter, I offer a diachronic analysis supporting hypothesis (A) above on the nature
of pronominal NSs, that is, that they are real pronouns, but silent at PF (null or deleted
elements). Being silent at PF, they give rise to a gap in the input a learner parses, and enter into
a paradigmatic relation with other ‘nulls’ or possible interpretations of gaps, which learners
encounter in their Primary Linguistic Data (PLD).
The historical analysis offered here is based on the ‘contingent’ view on syntactic change
(Lightfoot 1991, 1999) and the notion of the inertia of syntax (Longobardi 2001). Such a
methodological viewpoint implies that, in regular conditions, learners should easily acquire the
‘correct’ null category corresponding to a specific gap, given the syntactic and semantic
properties displayed by that gap. The opposite process, that is, shifting diachronically from one
null category to another in a specific gap (i.e. acquiring a different null element with respect to
previous generations), can only happen for a very good reason. The conditions and motivations
3
for such a change in the case of the Russian null categories will be analysed in detail in this
chapter.
In this sense, I will first show that Old Russian was a consistent null-subject language, of
the type of modern Spanish or Italian, allowing referential NSs (pro) in embedded and non-
embedded contexts. After the consistent pro-drop character was lost for independent reasons in
Middle Russian, speakers were deprived of the possibility of parsing a pronominal referential
null category pro in the gap left by embedded subjects. Therefore, they had to resort to the other
null category available in the language in order to interpret those embedded silent subjects, that
is, PRO, understood as a trace of movement or a copy, assuming the so-called Movement
Theory of Control (Hornstein 1999; Boeckx et al. 2010a).1
If speakers had the diachronic (acquisitional) possibility of choosing between these two
alternatives to ‘fill in’ a subject gap, pro should include some pronominal or featural content,
that is be ‘something’ parsable, in order to be able to enter into competition for the relevant
position with other alternative null elements (copies or traces—PRO). Besides, the alternative
hypothesis, pro-drop as rich agreement (i.e. loss of pro-drop as impoverishment of personal
features), will be shown to be an unrelated phenomenon, both historically and synchronically:
(i) impoverishment cannot account for the change in non-finite clauses, which always lacked
personal morphology; (ii) person morphology on the verb was preserved in the course of the
history of Russian (except for the past forms), regardless of the licensing of null subjects; (iii)
even in the case of loss of personal morphology in past tense, it developed in an independent
way with respect to the change analysed in this chapter.
Other implications of this study are the following: partial NS-languages are not as
homogeneous as we would expect if they represent a particular setting of a consistent parameter;
pro-drop itself does not seem to be a homogeneous parameter, but a set of related properties
interacting with independent phenomena, such as control structures.
The null element pro corresponds to a weak pronoun in consistent pro-drop languages. In
this sense, the Avoid Pronoun principle (Chomsky 1981; Rizzi 1997a), at least from a
diachronic point of view, can be reformulated from the point of view of the (non-)obligatory
character of pro, instead of its availability in a specific syntactic position. In simpler words,
1 For embedded subjects, we can assume any other of the alternative conceptual equivalents to traces, e.g. copies
or PRO. In this chapter, I will assume the Movement Theory of Control, as it accounts better for the diachronic
facts offered here, as we will see later on.
4
diachronically speaking, the pertinent question will not be whether pro is available or not in a
language, but whether pro is obligatory or not in a relevant context, as we will see.
In Section 7.2, I describe the conditions licensing NSs in Present-Day Russian (PDR) and
Old Russian (OR). In Section 7.3, I review the historical process of losing pro-drop in Middle
Russian. Section 7.4 evaluates the implications of all these diachronic and synchronic data for
a better understanding of the nature of pro. Finally, I provide some concluding remarks.
7.2. Conditions licensing Null Subjects in Present-Day Russian and Old
Russian
7.2.1. Null Subjects in Present-Day Russian
Traditionally, studies on Present-Day Russian (PDR) null subjects (NS) revolved around the
discussion about whether Russian is a canonical pro-drop language or a non-pro-drop language
(cf. Franks 1995; Bar-Shalom and Snyder 1997; Lindseth 1998; Demjjanow and Strigin 2000;
Perlmutter and Moore 2002; Müller 2006). In this chapter, I will assume the extended
characterization of PDR as a partial NS language; in fact, Russian is well-behaved with respect
to the basic properties acknowledged as typical of partial NS-languages, following Biberauer
et al. (2010) (cf. also McShane 2009, within a different perspective).
According to McShane (2009), the baseline realization option of pronominal subjects in
PDR is overt, that is, in neutral informational and discourse conditions, subjects are overt. But
subjects can be dropped under certain circumstances. In this section, I will detail the contexts
in which NSs are regularly used in PDR, together with their recoverability conditions. For ease
of exposition, I will list the conditions of licensing NSs in PDR by comparing them with the
usual characterization of partial NS-languages (in e.g. Holmberg et al. 2009):
(a) With generic or arbitrary reference: in impersonal finite clauses, unlike Finnish,
Marathi, or Brazilian Portuguese, the 3rd personal plural has a generic reading in Russian (1a–
b). Infinitive root (not embedded) impersonal sentences have an arbitrary reading (the so-called
‘arbitrary PRO’), as in (1c).
(1) a Tam e gotovjat znamenityj gruzinskij xleb.
there cook.3.PL famous Georgian bread
‘Traditional Georgian bread is baked there.’
b Zdes’ e ne kurjat.
here not smoke.3.PL
‘Smoking is not allowed here.’
5
c Kak krasivo [e exat’ po nočnoj trasse]!
how beautiful go.INF by night road
‘How beautiful is motorway driving at night!’
(b) In subjectless sentences, i.e. there are no overt expletives in PDR (2).
(2) (Veter / Zdes’ / V mašine) duet.
(wind / here / in car) blows
‘It is windy (here / inside the car).’
(c) In referential cases, only if they are ‘recoverable’. In PDR, we observe two splits in the
conditions licensing NSs:
(i) A split according to the personal feature: in PDR, the 1st and 2nd person (singular
and plural) are always recoverable from the discourse context, on a speaker-hearer basis
(3a–b) (cf. Holmberg et al. 2009, Sigurðsson 2011a),2 while 3rd person NS is licensed
by specific grammatical requirements, detailed in the next paragraph. Notice that
examples (3a–b) can be used in an out-of-the-blue context, and that using the 1st/2nd
personal overt pronoun is optional:
(3) a Vsem privet. (Ja) uže vernulas’!
all hello (I) already returned.F.SG
‘Hello everybody, I am already back!’
b Privet, kogda (ty) vernulas’?
hello when (you) returned.F.SG
‘Hello, when did you come back?’
(ii) A split according to the degree of clausal embedding, which takes place only in the
3rd person (singular and plural). According to Tsedryk (2013), 3rd person NSs are
allowed in:
– Root finite clauses, when they are bound topics, i.e. the antecedent is very close, usually an
immediately preceding definite NP (4a),3 and, crucially, licensing of an NS in this configuration
is blocked by any other element fronted to the left periphery of the clause (4b).
2 Taking into account what Holmberg et al. (2009: 61) report about Finnish and Marathi, a stylistic difference
emerges between PDR/Marathi vs. Finnish in this regard: Finnish allows dropping the 1st and 2nd person more
commonly in written and formal registers, while this kind of dropping in Marathi is more consistent in the spoken
language. In this sense, Russian patterns with Marathi, rather than Finnish.
3 Sentences including a 3rd person NS cannot be out-of-the-blue sentences, in contrast to the 1st/2nd person
examples in (3). If the antecedent is not included in the immediately preceding sentence, the NS is inserted anyway
in some sort of dialogue, sometimes even forming a ‘chain of NSs’ together with other co-referent null pronouns,
6
(4) a Ja tol’ko čto vstretil Lenui. Ona / ei/*j skazala, čto ix
I.NOM just met.M.SG Lena.ACC she said.F.SG. that their
otdel skoro zakrojut. (From Tsedryk 2013)
department soon will close
‘I have just met Lena. She said that their department will be closed soon.’
b Ja tol’ko čto vstretil Lenui. Čto onai / *ei delaet na kuxne?
I.NOM just met.M.SG Lena.ACC what she does.3.SG in kitchen
‘I have just met Lena. What is she doing in the kitchen?’ (From Tsedryk 2013)
- Embedded finite clauses, under Obligatory Control (OC),4 i.e. only when they have a very
local controlling subject antecedent (5a-d) (cf. also Landau 2004).5
(5) a Lenai boitsja, čto Majjaj dumaet, čto (onai/j) / ej/*i ne sdast èkzamen.
Lena fears that Maia thinks that she not will pass.3.SG exam
‘Lena is afraid that Maia thinks that she (=Maia) will not pass the exam.’
b Majjai boitsja, čto roditeli dumajut, čto onai / *ei ne pridët vovremja.
Maia fears that parents think that she not will come.3.SG on time
i.e. all ‘hanging’ from the same overt referent, and without any other possible referent intervening between them.
Here is an example (I thank an anonymous reviewer for pointing out this fact to me):
(i) – Gde Lenai?
where Lena
– Ešče ei ne prišla.
yet not came
– Kogda ei pridët, pust’ ei pozvonit mne na sotovyj.
when will-come let will-call me to mobile
‘– Where’s Lena?—She hasn’t arrived yet.—When she arrives, she should call me.’
An account of Russian NSs in root finite clauses is beyond the scope of this chapter, but cf. Tsedryk (2013), where
it is argued that such elements are not really NSs but instances of null topics or topic drop à la Sigurðsson (2011a),
which is consistent with the existence of a ‘chain of nulls’ like those we find in Russian.
4 The rest of the OC properties, according to the classic description by Hornstein (1999), are exemplified for PDR
finite embedded clauses in Tsedryk (2012), namely, the ban on split antecedents, sloppy interpretation under
ellipsis, and only de se reading.
5 In this respect, Russian seems similar to other partial NS-languages, such as Finnish, Brazilian Portuguese, and
Marathi, according to Modesto’s (2008), Nunes’s (2010) and Holmberg et al.’s (2009) descriptions. A closer look
at the crucial structures reveals several differences between these languages: for example, spoken Russian strictly
applies the constraints illustrated in examples (5a–d), while spoken Finnish would be more permissive in using
NSs in embedded contexts other than OC, at least according to Frascarelli’s (2014) survey. However, a cross-
linguistic comparison is beyond the scope of this chapter.
7
‘Maia is afraid that her parents think that she will not come on time.’
c Razgovor s Majeji dokazal, čto onai / *ei terjaet svoe vremja.
conversation with Maia showed that she loses.3.SG her time
‘My conversation with Maia proved that she was wasting her time.’
d Jai ubedil Majjuj, čtoby onaj / *ej prišla.
I convinced.M.SG. Maia.ACC that she come.F.SG
‘I convinced Maia to come.’
The previous examples illustrate the most salient properties of NSs in PDR. The examples in
(4) correspond to NSs in root contexts; in (4a), the only possible antecedent for the NS in the
second root clause is Lena, which is overtly expressed in the previous clause. Example (4b)
shows that the fronted wh-element čto ‘what’ blocks the binding relation available in (4a).6 The
examples in (5) illustrate the Obligatory Control (OC) pattern of NSs in finite embedded
clauses: (5a) shows that an NS is licensed only when it is co-referent with the closest potential
antecedent; (5b) shows that an NS is excluded when the closest potential antecedent is not local;
(5c) illustrates the fact that an NS is excluded when its potential antecedent does not c-command
it, and (5d) that the antecedent must be a subject of the root clause.
Crucially, unlike in other languages, these recoverability conditions are completely equal
whether the verbal form displays overt personal features (in present and future tenses, with
overt personal morphology) or not (with impoverished past l-forms, which lack personal
morphology). The examples in (6a–b) are the morphologically ‘rich’ counterparts of (3a–b),
while those in (7a–c) illustrate the morphologically ‘impoverished’ counterparts of (5a–c). Note
that the verb in (5d) does not have an equivalent with overt personal morphology, as it is an
instance of the Slavic equivalent of subjunctive clauses, which obligatorily includes a defective
l-form.
(6) a (Ja) edu v Nižnyj, e vernus’ zavtra.
(I) go1.SG to Nizhny will return.1.SG tomorrow
‘I am going to Nizhny Novgorod, and will be back tomorrow’
b Privet, kogda (ty) vernëš’sja?
hello when (you) will return.2.SG
‘Hello, when are you coming back?’
6 I illustrate the blocking effects with one of Tsedryk’s (2013) examples, namely, blocking by a fronted wh-
element, but any other fronted element gives rise to the same effect (fronted topics or foci, fronted personal
pronouns, fronted PPs, etc.).
8
(7) a Lenai boitsja, čto Majjaj dumaet, čto onai/j / ej/*i ne sdala èkzamen.
Lena fears that Maia thinks that she not passed.F.SG exam
‘Lena is afraid that Maia thinks that she (=Maia) did not pass the exam.’
b Majjai boitsja, čto roditeli dumajut, čto onai / *ei ne vernulas’ vovremja.
Maia fears that parents think that she not came. F.SG on time
‘Maia is afraid that her parents think that she did not return on time.’
c Razgovor s Majeji dokazal, čto onai / *ei poterjala svoe vremja.
conversation with Maia showed that she lose. F.SG her time
‘My conversation with Maia proved that she had wasted her time.’
As shown in Madariaga (2011), infinitive embedded clauses also display OC in PDR; cf. a
couple of properties in the examples in (8): (8a) illustrates the property that non-finite embedded
NSs must have an overt local antecedent; (8b) evidences a sloppy interpretation under ellipsis:7
(8) a * Vanjai dumal, čto ožidalos’ [ei/j poznat’ sebja].
Vanya though that was expected know.INF himself
Intended: ‘Vanya thought that he was expected to know himself.’
b Vanjai xočet [ei poznat’ sebja] i Anja tože.
Vanya wants know. INF himself and Anya too
‘Vanya wants to know himself and Anya does too (=Anya know herself).’
PDR features a typical property of partial NS languages: even in those configurations where an
NS is available, its use is optional, in the sense that (i) it can always be replaced by an overt
pronoun in colloquial speech, and (ii) it is certainly replaced by an overt variant in neutral and
literary speech. On the other hand, NSs are almost never obligatory in PDR (as we will see later,
the reverse situation is characteristic of ‘consistent’ NS-languages).
A very intuitive generalization by McShane (2009) is a good starting point to grasp the
pragmatic aspects of the distribution of overt versus null subjects in PDR: even in cases where
an NS is in principle allowed, the more complex the structure or/and the more ‘heavy’ the
intermediate material between the subject and its antecedent, the more favoured an overt subject
is. This generalization covers the following observations:
(i) Even in those cases where an NS is licensed or even preferred in colloquial language, the
overt variant is also licit (9a); cf. also examples (3a–b, 4a, 6a–b, 7a).
7 Other OC properties of PDR infinitive embedded clauses (the ban on split antecedents, the need for a c-
commanding antecedent, etc.) are described in Madariaga (2011).
9
(ii) The only clear cases when an NS is compulsory can be characterized as instances of
coordination of syntactic components lower than TP;8 cf. example (9b) below or (6a) above,
both illustrating two coordinated VPs (PDR does not display V-to-T movement; cf. Bailyn 2012;
Tsedryk 2012).
(iii) The ‘heaviness’ of the material between the subject and its antecedent can render the
reverse effect: in the second conjunct of example (9c), the null variant is highly dispreferred.
Allegedly, the locus of the coordination of these components must be higher here than in (9b).9
(9) a Oni sel potomu, čto oni / ei/*j ustal. (McShane 2009: 119)
he sat because that he was tired
‘He sat down because he was tired.’
b Oni [sel] i ei / *oni [načal est’]. (McShane 2009: 118)
he sat and he started eat
‘He sat down and started to eat.’
(McShane 2009: 119)
c Oni sel rjadom so mnoj na divane i posle dolgogo molčanija, oni /*% ei načal est’.
he sat near with me on sofa and after long silence he started eat
‘He sat down near me on the sofa and, after a long silence, he started to eat.’
7.2.2. Null subjects in Old Russian
8 Root finite clauses used as an answer to ‘yes/no’ questions display a similar effect; here, an NS can be obligatory,
as shown in (i), depending on the informational structure of the statement, just as in Finnish, Brazilian Portuguese,
and Marathi (Holmberg et al. 2009: 65–7). Conversely, in answers to wh-questions, the effect is the reverse (ii):
(i) – Prišel Saša? (‘Did Sasha arrive?’)
– *On / e prišel.
he came (‘Yes (he did)’)
(ii) – Kogda Saša prišel? (‘When did Sasha arrive?’)
– On / *e prišel včera.
he came yesterday (‘He arrived yesterday.’)
However, as shown by Holmberg et al. (2009), these are independent pragmatic effects and irrelevant for the issue
of pro-drop. Therefore, I will not discuss them here.
9 McShane (2009) gives (9c) as ungrammatical. However, some speakers accept as possible the null variant in
example (9c), while others find it much worse than the overt variant, especially compared to ‘simpler’ examples
such as (9a–b). In any case, (9c) stands for a configuration where NSs are grammatically licensed; it is not therefore
surprising that the interference of unclear pragmatic considerations does not trigger the same effect in all the
speakers.
10
In this section, I will review the use of NSs in Old Russian, highlighting the contrast between
their properties then and nowadays. Unlike PDR, Old Russian (OR) is commonly
acknowledged to display the properties of consistent NS-languages, according to Biberauer et
al.’s (2010) classification. The typical contexts where overt subjects are used in OR are
comparable to those in Spanish, Italian, and other consistent NS-languages, more specifically,
in contrastive or emphatic configurations (cf. Rizzi 1986a; Roberts and Holmberg 2010).10
Otherwise, an NS is mandatorily used, in accordance with the Avoid Pronoun principle
(Chomsky 1981; Rizzi 1997a, etc.).
Likewise in OR, an NS was used by default in ‘non-discourse-related’ contexts, while overt
pronouns were always emphatic or contrastive, according to traditional works, such as Lomtev
(1956: 35ff), Borkovskij (1968: 17ff, 1978: 10ff, 187ff), and more recent works, especially
Meyer (2011, and references therein).
Let us consider a parallel example of the occurrence of an NS in OR11 and Spanish (10a–
b), on the one hand, and its PDR equivalent, on the other hand (10c). These examples illustrate
root sentences with a 3rd-person contextual antecedent (Jesus), which is not mentioned overtly
immediately before the example, but is easily inferred from the previous text.12 As shown in
(10a–b), in OR and its Spanish translation a pronominal subject is not and cannot be overt here.
10 In this chapter, I illustrate emphatic configurations with the help of contrastive topics but, obviously, foci are
also always overt. For example, to the question ‘Who helped her doing her homework?’, we answer with a
preverbal or post-verbal stressed overt subject in both Russian (JA jej pomog ‘I helped her’) and Spanish (YO le he
ayudado ‘I helped her’ or Le he ayudado YO lit. ‘her helped I’).
11 References of the primary sources along this chapter: Deed: Akademija nauk (1836); Ivan the Terrible's epistles:
Lixačev, Lur’e, and Adrianova-Peretc (1951); Laurentian Chronicle: Karskij (2001 [11926–28]); Mstislav’s letter:
Obnorskij and Barxudarov (1952); 1st Novgorod Chronicle: Nasonov (2000 [11950]); Novgorod letters: from the
web site: gramoty.ru, based on: Zaliznjak (2004); Sermon on Law and Grace: Sreznevskij (1893); Sinodal
Patericon: Golyščenko and Dubrovina (1967); Suzdal Chronicle: Karskij (2001 [11926–28]).
12 The entire passage of the Laurentian Chronicle, where example (10a) can be found, is about Jesus, who qualifies
here as the crucial Aboutness-shift topic, in Frascarelli’s (2007) terms. The previous lines to example (10a) are
translated as follows: ‘Pilatus condemned Jesus to crucifixion and they crucified him (=Jesus) (…) The Hebrew
people stationed guards because they said that the disciples might abduct him (=Jesus). But he (=Jesus) resurrected
on the third day, and appeared to his disciples. pro (=Jesus) rose from the dead, and said to his disciples: go to
every village and teach all the people the baptism in the name of the Father, the Son and the Holy Spirit. pro
(=Jesus) stayed with them 40 days, appearing to them after his resurrection. When 40 days had passed, pro (=Jesus)
asked them to go to the Mount of Olives and …’
11
In the PDR translation in (10c) we observe the reverse effect, that is, an NS is proscribed in this
context:
(OR: Laurentian Chronicle 35v)
(10) a ei Povelě imъ iti v goru Elevonьskuju i tu ei javisja imъ.
commanded.3SG them go to mount Eleon and there appeared them
‘He commanded them to go to the Mount of Olives and there he appeared to them.’
b e / *él les ordenó ir al Monte de los Olivos y e / *él allí se
he them commanded.3SG go to Mount of the Olives and he there REFL
les apareció. (Spanish)
them appeared.3SG
c Oni / *ei prikazal im pojti v Masličnuju goru i tam oni / ei
he commanded.M.SG them go to Olive Mount and there he
javilsja k nim. (PDR)
appeared.M.SG to them
I will not discuss here elaborated theories on the different types of topics (for this, cf. Frascarelli
and Hinterhölzl 2007). I will just suggest that the interpretation of NSs as Aboutness-topics in
consistent NS-languages seems also to hold in OR (contrasting with the interpretation of overt
pronouns as contrastive topics or emphatic elements). Additionally, consider the following
example:
(11) Toi že vesne oženisja knjazь Mьstislavъi Novegorodě (…). I potomъ pozvaša
this PRT spring married prince Mstislav in Novgorod and then called
i rostovьci kъ sobe, i ei ide Rostovu sъ družinoju svoeju,
him Rostovians to themselves and went to Rostov with army his
a ei synъ ostavi vъ Novegorodě, i ei pride Rostovu. I vъ to vrěmja
and son left in Novgorod and went to Rostov and at this time
umьrlъ bjaše Mixalko. I ei poide sъ rostovьci i sъ suždalьci
died was Mikhail and went with Rostovians and with Suzdalians
kъ Volodimirju… (OR, 1st Novgorod Chronicle, 40-40v)
to Vladimir
‘In the spring of that year, the prince Mstislav got married in Novgorod (to Yakun
Miroslavich’s daughter). Then, the people of Rostov called him (= Mstislav) for help,
and pro (= Mstislav) went to Rostov with his army and pro (= Mstislav) left his son
in Novgorod, and pro (= Mstislav) attacked Rostov. At that time, Mikhail had already
12
died. And pro (= Mstislav) attacked the town of Vladimir together with the people of
Rostov and Suzdal.’
The whole text is about Mstislav, prince of Novgorod, and regardless of other grammatical
subjects and possible antecedents that are introduced in the text as secondary (the people from
Rostov, Mstislav’s son, and Mikhail, the prince of Vladimir-Suzdal), every NS in the fragment
refers to Mstislav.
On the other hand, in OR, overt pronominal subjects were used in contrastive or emphatic
contexts, frequently followed by a series of particles with the function of a discourse marker:
the contrastive enclitic particle že, and/or the emphatic proclitic particles bo, i, da, and i
(Borkovskij 1978: 193). A parallel example of a typical contrastive topic in Spanish and OR is
given in (12); the OR subject in (12a) is followed by the contrastive enclitic particle že:
(12) a Reč(e) že Volodimerъ čto radi ot ženy rodisja (…) On že
said.3.SG PRT Vladimir what because from woman was born he PRT
reč(e) emu sego rad(i) poneže isperva rodъ č(e)l(o)v(e)č(e)skii
said.3SG him this because because firstly kind human
ženoju sgrěši. (OR: Laurentian Chronicle 35v)
woman committed sin
‘Then, Vladimir asked (the philosopher) why he was born from a woman,
(crucified on a tree and christened with water), and he (=the philosopher) told
him (=Vladimir) that because mankind committed sin for the first time through
a woman...’
b Juani dijo que lej ayudaría pero élj / *e lei dijo que no. (Spanish)
Juan said that him help but he him said that no
‘Juan told him that he would help him, but he told him not to.’
7.2.3. Interim summary: null subjects in Old Russian and Present-day
Russian
Summarizing so far, we have established that PDR is a typical partial NS-language, with the
particularity of being very permissive in licensing 1st and 2nd person NSs, but more restrictive
in licensing 3rd person referential NSs (in particular, when they are bound topics in root finite
contexts, and under Obligatory Control in finite and non-finite embedded contexts).
On the contrary, OR was a consistent NS-language, like modern Spanish or Italian. We can
assume that NSs were obligatory in OR when they introduced an element present in the
13
discourse (an Aboutness-shift topic, following Frascarelli 2007) or, at least, unless there was
some contrastive or emphatic informative feature in C. Likewise, an overt pronoun was used in
case it was a contrastive topic or an emphatic element, such as a focus.
7.3. The development of null subjects in Russian
7.3.1. Change of the pro-drop character of Russian in root or non-embedded
contexts
The change in the availability of null subjects did not pass unnoticed in traditional works on
Russian syntax (Potebnja 1958 [1888]; Istrina 1923; Borkovskij 1949, 1968, 1978; Lomtev
1956). These authors highlighted the ‘emphatic’ character of overt pronouns in early Russian,
which shifted into ‘non-emphatic’ in later Russian, and characterized this development as the
loss of the ‘logical stress’ of OR overt pronouns. Other studies offer more detailed descriptions
of this change. Nevertheless, accounting for the development from consistent into partial pro-
drop is beyond the scope of this chapter because the stage in which the consistent NS-character
of Russian got lost is just the starting point for the facts I will analyse here (cf. § 7.4).
Thus, in this section I will just briefly review the crucial steps of the process of losing the
availability of pro in ‘non-emphatic’ subject position in non-embedded clauses. This process
was described in Černyx (1954), Avanesov and Ivanov (1982), Ivanov (1990), Lindseth (1998),
and Meyer (2011):
(1) Old Church Slavonic (the oldest attested Slavic language) and, to different degrees, also
other early Slavic languages, displayed a whole range of tensed verbal forms (present, perfect,
aorist, imperfect, pluperfect, etc.).
(2) The perfect form was an analytic construction including the so-called l-participle (with
gender and number morphology) and a tensed (present) form of the copula byti ‘to be’, which
functioned as an auxiliary and displayed person and number morphology, as illustrated in (13a–
14
b).13 This perfective construction prevailed as the only past form in Russian, replacing other
past forms by the fourteenth century (aorist, imperfect, pluperfect).14
(Laurentian Chronicle, 14v)
(13) a Počto ei ideši opjatь, ei poimalъ esi vsju danь.
what for go.2.SG again took.M.SG AUX.2.SG whole tax
‘Why did you come back? You collected the tribute already.’
b Vižь sego ty ježe esi xotělъ. (Laurentian Chronicle, 23v)
look this you if AUX.2.SG wanted.M.SG
‘See whether this was what you (i.e. but not me) wanted.’
(3) Overt subjects were used in emphatic and contrastive contexts (cf. §b7.2.2) in the form of
regular personal pronouns in the case of the 1st and 2nd person. Dedicated personal pronouns
for the 3rd person did not exist in early Slavic; a series of demonstrative pronouns was used
instead. The Russian 3rd person pronoun emerged in Middle Russian from a reanalysis of two
of those demonstratives, namely, onъ, ona, ono ‘that’ (third degree of distance) for the
nominative case, and i, ja, je ‘this, that’ (neutral degree of distance) for oblique cases
(Borkovskij and Kuznecov 1964).15
(4) Overt auxiliaries in perfect (past) constructions were progressively lost. In many cases, an
overt NP or an overt pronoun was used instead.16 The loss of the 3rd person auxiliary (14a) was
13 I provide two examples from the same chronicle: (13a) displays two NSs with a perfect form (l-form + auxiliary),
while (13b) illustrates an overt emphatic pronominal subject with a perfect form (again, l-form + auxiliary). The
context of example (13b) is the following: Sveneld wants to avenge his son, killed by Oleg, and convinces
Yaropolk, Oleg’s brother, to conquer Oleg’s lands. While Yaropolk and Sveneld attack the city, nobody notices
that Oleg falls from the overcrowded drawbridge and dies. Yaropolk takes the city and looks for his brother; when
he finds Oleg’s body, Yaropolk pronounces the sentence in example (13b), blaming Sveneld for his brother’s
death.
14 I will not consider here the development of the present-tensed copula in its purely copular use, but only as an
auxiliary verb combined with the l-participle forms (these two uses display homonymic forms) but, incidentally,
the loss of the present-tensed copula in both these uses took place quite simultaneously (Lomtev 1956: 35ff).
15 According to Borkovskij (1978: 11, 312), the first personal uses of the demonstrative pronoun date from the
thirteenth century, but it was not until the fifteenth to seventeenth centuries that it became more widespread as a
personal pronoun only.
16 Jung (2014) analyses this correlation in terms of diachronic loss of redundant agreement. Following Roberts’
(2010a) account of the distribution of a D-feature on T as the source of the pro-drop parametric variation in
languages, Jung argues that OR, as a well-behaved consistent NS-language, displayed a D-feature on T, as well as
V-to-T movement. This situation changed later on, due to the loss of temporal distinctions in Russian (in favour
of a verbal system based on aspectual distinctions). Simplifying Jung’s account, at that stage, V-to-T movement
15
a generalized phenomenon from the very first Russian texts, while 1st and 2nd person
auxiliaries got lost later. Overt pronouns were emphatic in Old Russian (14b):
(14) a O(tь)cь moi umerlъ. (Laurentian Chronicle, 48)
father my died.M.SG
‘My father died.’
b Ne jazъ bo počalъ bratьju biti no onъ. (Laurentian Chronicle, 24)
not I PRT started.M.SG brothers hit but he
‘It was not me, but he, who started to attack our brothers.’
(5) Personal morphology in present, future and imperative forms was never lost nor suffered
phonological attrition; nevertheless, the insertion of ‘non-emphatic’ overt pronouns in the place
of NSs affected them to the same extent and at the same rate as defective or ‘impoverished’
past forms (Lomtev 1956; Meyer 2009).
(6) The loss of obligatory NSs (the facultative introduction of a weak pronoun) denoting the 1st
and 2nd person correlated chronologically with the loss of 1st and 2nd person auxiliaries; in the
case of the 3rd person, however, the auxiliary was lost long before overt 3rd person pronouns
generalized as weak or ‘non-emphatic’ (Meyer 2009, 2011).
The development of the pro-drop phenomenon in Russian reveals a clear asymmetry between
the 3rd person and 1st/2nd persons: even if 3rd person auxiliaries were lost long before 1st/2nd
person auxiliaries, overt weak pronouns for the 1st/2nd person started to be used before weak
pronouns of 3rd person. Lindseth (1998) explains this asymmetry in the following way: 3rd-
person auxiliaries could be more easily dispensed with because, unlike the 1st/2nd person,
where there was no alternative way to realize an overt subject other than using a personal
pronoun, in the 3rd person, any overt NP subject (not just an overt pronoun) could cancel out
the need for a personal auxiliary. The larger number of NP subjects compared to (still emphatic)
1st and 2nd person overt pronouns triggered a higher frequency of instances licensing the lack
of a copular auxiliary in 3rd person perfect constructions, but not in the 1st and 2nd person.
A possible initial environment for the reanalysis of 1st and 2nd person pronouns as weak
pronouns could be certain fixed formulae, typically found in letters dating from the earliest
Russian period (both in Northern and Central-Southern OR letters). According to Borkovskij
(1968), the first overt pronominal subjects which could be interpreted as ‘non-stressed’ (just
stylistically motivated pronouns, according to this author) were those included in formulae such
was lost; thus, pronouns started to satisfy the EPP requirement on T, leading to the eventual loss of overt auxiliaries
in the past tense, as they were redundant with respect to the newly introduced overt weak pronouns.
16
as a se azъ / ja ‘and PRT I’, a azъ / ja ‘and I’, and the like. Such formulae gave rise to the earliest
cases of ‘ambiguous’ or ‘unclear’ interpretation of overt pronouns in the thirteenth to fourteenth
centuries. Example (15a), from an early letter, illustrates the fact that an overt pronoun included
in such a formula was interpreted as was usual in OR, that is, as discursively motivated
(contrastive). However, later instances of overt pronouns within the same formulae seem just
stylistically motivated, as illustrated in examples (15b) and (15c), from the thirteenth and
fourteenth centuries, respectively:
(15) a A se ja vsevolodъ dalъ esmь bljudo serebrьno vъ I grьvnъ
and PRT I Vsevolod given.M.SG AUX.1SG. dish silver in 8 grivnas
serebra s[vja]t[o]mu že georgievi. (OR: Mstislav’s letter, year 1130)
silver saint PRT George
‘And I, Vsevolod, gave Saint George a silver dish worth 8 silver grivnas.’
(OR: Novgorod letter #147)
b A pocta tobь bude gъzhe a azo tobe klanjajusja.
and honour you will be good and I you bow.1SG
‘And you will preserve your good reputation. I bow to you (= I say goodbye).’
(OR: Novgorod letter #222)
c Ožь li e pravo zapirajutsь(ja) a ja daju knjažju dьcьskamu grivnu
if PRT right are obstinate and I give.1.SG of prince children’s grivna
sьrьbra.
silver
‘If they do not confess, I will pay a silver grivna to the prince’s administrator.’
Example (15a) is from a very early letter (around the year 1130), and illustrates a contrastive
topic: the initial part of the letter was written by Prince Mstislav, and the sentence in example
(15a) introduces a second author, his son Vsevolod, who writes the final part of the letter. Here,
the overt pronoun co-occurs with an overt auxiliary.
In contrast, the expression a azъ in example (15b) heads the final sentence (a closing
formula) of the corresponding letter, introducing what can qualify as the general topic of the
letter, that is, its writer (first person). Given the fact that OR was a consistent NS-language, we
would expect such a subject to be null; however, the stylistic value of the corresponding fixed
formula renders it as overt. The same effect is observed in example (15c), where the overt
subject within the formula a ja ‘and I’ is not a clear contrastive topic but rather a common topic:
although there are a few words missing in the birch bark, the letter introduces the first person
and his whereabouts as the main topic of the text. In both cases, the equivalent in a consistent
17
NS-language (e.g. Spanish) would be an NS: e Te digo adiós or e Me despido in the case of
(15b), and Si ej (=ellos) no quieren confesar, ei (=yo) pagaré una grivna al administrador for
(15c).
Further, co-occurrence of an overt pronoun and an auxiliary displaying personal
morphology, as in (15a), was not an extended pattern in OR perfect-past tenses. In fact,
according to Zaliznjak (2004, 2008), in Northern OR the only patterns found were (i) ‘overt
pronoun or NP + l-participle’ (ja dalъ ‘I gave’) and (ii) ‘l-participle + auxiliary’ (esmь dalъ
‘AUX.1SG. gave’); a ‘redundant’ pattern (iii) ‘overt pronoun or NP + l-participle + auxiliary’
was virtually unattested. According to Lomtev (1956), in Central and Southern OR, pattern (iii)
was available, although the norm was pattern (i).
To sum up, on the one hand, 3rd person auxiliaries were lost before 1st and 2nd person
auxiliaries, maybe just because overt NPs were much more frequent than emphatic 1st/2nd
person pronouns; hence, they cancelled out more easily the absence of the corresponding overt
auxiliary in past-tensed forms. On the other hand, the data presented so far suggest that 1st/2nd
person emphatic pronouns were more readily reanalysed as weak pronouns than 3rd person onъ,
ona, ono, because of (i) the existence of fixed formulae including 1st/2nd person pronouns (not
3rd person pronouns), which raised the possibility of reanalysing them as weak pronouns, and
(ii) again, the low frequency of 3rd person overt (emphatic) onъ, ona, ono, in comparison with
a much higher frequency of overt NP subjects and other competing pronouns (a whole range of
demonstrative pronouns other than onъ, ona, ono). An interesting observation is that this
asymmetry in the historical development of Russian pro-drop correlates with the split according
to person features described for the synchronic facts in § 7.2.1.
The initial extension of non-stressed non-emphatic (weak) overt pronouns in the 1st/2nd
person in OR can be posited by the fifteenth century, roughly correlating with the loss of the
auxiliary verb displaying 1st/2nd personal morphology in perfect tenses. Third person weak
pronouns emerged later (sixteenth to seventeenth centuries), probably due to the fact that overt
NPs, as well as a variety of demonstratives, fulfilled the function of weak pronouns.17 In the
case of non-past tenses, personal morphology was never lost nor underwent phonological
attrition, but the introduction of overt weak pronouns in these configurations proceeds in the
same way as in constructions including past (impoverished) forms.
17 Meyer (2009, 2011) offers later dates than traditional historical grammars, as he focuses on the decline of the
statistical preservation of NSs in referential contexts (not on the loss of their obligatory o ‘default’ use). Thus, he
posits the loss of 3rd person pro-drop by the seventeenth to eighteenth centuries.
18
In § 7.4, I will argue that the last stage of the historical change outlined in this section was
the relevant cue that led learners to reanalyse NSs in embedded clauses in Russian. In other
words, the loss of the consistent NS-character was the initial point of the reorganization of
crucial aspects of the system of embedding in the Russian language. Therefore, the period I
focus on is not the moment when pro-drop was definitely lost in OR (perhaps as late as the
eighteenth century, according to Meyer 2009, 2011), but the moment when speakers did not
need to parse every regular topic as an NS, and started to have the possibility of positing an
overt pronoun instead.
Before discussing embedded clauses, I need to make a final remark on non-embedded
constructions; namely, that both early Slavic and PDR display overt dative subjects in root
infinitive clauses, as shown in (16):
(16) a Bratu tvojemu Kyeva ne uderžati. (OR: Suzdal Chronicle 108b)
brother.DAT your.DAT Kiev not maintain.INF
‘Your brother is not able to keep Kiev.’
b Čto našej sem’je / nam eščë bylo delat’? (PDR)
what our family.DAT / us.DAT yet was do.INF
‘What else could we / our family do?’
I will not discuss root infinitive clauses in detail, as their syntax did not change much during
the history of Russian; for a complete study of root infinitive clauses in Slavic, the reader is
referred to Wiemer (2017). The important point to make here is that the availability of
referential NSs in root infinitive clauses experiences more severe restrictions than in root finite
clauses: they are reduced to instances of the so-called ‘arbitrary PRO’, as in example (1c) in §
7.2.1. This is allegedly due to their special status: root infinitive clauses in Russian were/are
interpreted as modal, as shown in the English translation of (16), that is, the infinitive has an
intrinsic modal value (cf. Fortuin 2005; Kotin 2012) and its dative subject is comparable to
regular dative subjects, usually experiencers, in finite clauses (Bailyn 2012: 171). For
independent reasons, explained in detail in Tsedryk (2013), we know that dative subjects in
Russian cannot be silenced ((17a)) even in cases when regular nominative subjects can ((17b)):
(17) a Lenai nadeetsja čto ej / *ei budet teplo. (PDR: from Tsedryk 2013)
Lena.NOM hopes that her.DAT will be warm
‘Lena hopes that she will be warm.’
b Lenai nadeetsja čto ona / ei budet znamenita.
Lena.NOM hopes that she.NOM will be famous
‘Lena hopes that she will be famous.’
19
The issue of root infinitive clauses will play a role in later discussion in this chapter. For now,
it is crucial to keep in mind (i) that dative subjects in root infinitive clauses, unlike embedded
non-finite contexts, were preserved during the whole history of Russian (the implications of
this fact will be addressed in §§ 7.4 and 7.5); and (ii) the role of case marking of embedded
finite NSs shown in example (17), that is the fact that they must correspond to a nominative
subject and that they must also have a nominative subject antecedent18 (to be addressed in §
7.3.2).
7.3.2. The development of null subjects in embedded structures
So far I have reviewed the loss of pro-drop in root contexts, and I will now outline the change
processes that affected NSs in embedded constructions in Russian. We will see that Old Russian
allowed null and overt subjects in finite and non-finite embedded contexts. Together with the
loss of pro-drop, embedded finite clauses preserved NSs only in instances of Obligatory Control,
and all embedded infinitive clauses became instances of Obligatory Control as well.
In early Slavic, not only NSs, but also overt subjects were licensed in every embedded
context, both finite and non-finite.19 In the case of non-finite clauses, an overt NP or pronoun
18 It is not just that an embedded finite NS cannot correspond to a dative experiencer subject; an embedded finite
NS cannot have a dative experiencer antecedent either:
(i) * Emui stalo stydno, čto ei sdelal ètu ošibku. (From Tsedryk 2012)
him.DAT became ashamed that made this mistake
‘He felt ashamed that he made this mistake.’
19 It is a well-known fact that all early Indo-European languages, including early Slavic, made widespread use of
a sort of juxtaposition as a strategy of embedding which, to our modern ears, sounds as if there were no
subordination at all. But they were not completely independent clauses either (there was no punctuation marking
or anything signalling e.g. an intonational break). This was typical in the case of finite and non-finite
complementation of jussive and declarative verbs, absolute constructions, and relative clauses. Sometimes it is
called semantic subordination or subordination of events, which has been shown to lack the ‘modern’ syntactic
subordinative properties that developed in later Indo-European languages. More specifically, most authors assume
an adjoined structure for this kind of ‘subordination’ (Andersen 1970; Garrett 1994; Kiparsky 1995; Keydana
2013, and references therein; cf. Madariaga 2014 in the particular case of East Slavic); the sequence of changes in
this development has been outlined in Disterheft (1980, 1997), specifically in the case of infinitive clauses. In the
case of the structures analysed in this chapter (infinitive clauses associated to jussive and declarative ‘matrix’
verbs), this development from bare semantic subordination into fully syntactic embedding (from juxtaposed /
adjoined into complement of a matrix verb) can be formalized as a change from no control into OC (cf. § 7.4).
Note, however, that I remain agnostic about the possibility of the existence of control in OR in other instances,
most likely in infinitive clauses and as complements of volitional verbs with a coreferent subject.
20
encoded with dative case could be the subject of an infinitive or a participial verbal form.
Infinitive and participial (the so-called absolute) constructions including an oblique subject
were widespread in all early Indo-European languages. As described in Madariaga (2015), the
distribution of non-finite constructions as mechanisms for the embedding of events in early
Slavic was roughly the following:
(i) The structure ‘(overt dative subject) + infinitive’, cf. (18a), could serve as a complement to
declarative, jussive, and volitional verbs, a complement to nominal elements or used as a
purpose clause;
(ii) Absolute constructions (‘dative subject + dative participle’), illustrated in (18b), performed
the function of other adverbial clauses (cf. the range of adverbial uses of absolute constructions
in Andersen 1970).
(OR: Sermon on law and grace)
(18) a A v tobĕ estь vlastь [ili žiti nam ili umrĕti].
and in you is power or live.INF us.DAT or die.INF
‘And you have the power over whether we live or die.’
(OR: Laurentian Chronicle, 3v)
b ei Divno vidĕxъ slovensьkuju zemlju [iduči mii sĕmo].
fantastically saw Slavic land come.PTCP.DAT me.DAT here
‘I had a beautiful sight of the Slavic lands while I was coming here.’
The absolute constructions, such as (18b), were lost at an early point in OR. The infinitive
structures were preserved, but underwent significant changes, related to the development of
NSs, as I will show in § 7.4.
Focusing on the development of infinitive structures, we observe that in early Slavic a wide
range of subjects was licensed in these clauses: (i) overt NPs, as in (19a); (ii) overt (emphatic,
contrastive) pronouns, illustrated in (19b);20 and (iii) null subjects, like the one in (19c). The
subjects illustrated in (19a–b) are instances of post-verbal dative subjects (existential and
unaccusative, respectively); preverbal, agentive, and unergative overt dative subjects were also
available in early Slavic (see Madariaga 2015 for a complete classification and more examples).
20 When an overt pronoun was used in infinitive embedded contexts, we know that it was emphatic not only due
to its interpretation but also according to its form. Until the thirteenth to fourteenth centuries there were two series
of accusative and dative pronouns in Russian: clitics (dative clitic series: 1sg. mi, 2sg. ti, 1pl. ny, 2pl. vy, 1du. na,
2du. va) and strong pronouns (dative strong series: 1sg. mъne, 2sg. tebě, 1pl. namъ, 2pl. vamъ, 1du. nama, 2du.
vama) (Zaliznjak 2008). Dative clitic pronouns are in decline in this period but the fact is that the overt pronouns
used in the constructions reviewed here are always of the strong type.
21
In addition, any of these subjects could have an associated dative secondary predicate, as in the
participle izbavlenъ in (19c) or the floating quantifier oba in (19b).21
(OR: 1st Novgorod Chronicle, 98)
(19) a ei I slyšaše [byti stuku i gromu veliku].
and heard.3SG. be.INF noise.DAT and thunder.DAT big.DAT
‘And he heard that there was a noise and a big thunderclap.’
(OR: Laurentian Chronicle, 170b)
b [Ty so mnoju]i cělovalъ kr(e)stъ [xoditi namai po odinoj dumě oběma].
you with me kissed cross go.INF we.DAT by one decision both.DAT
‘You and me swore that we both would do it the same way.’
(OR: Laurentian Chronicle, 71b)
c Molisja [za mjai] otče čestnyj [ei izbavlenu byti ot seti neprijazniny].
pray for me father honorable saved.DAT be.INF from this devilment
‘Honourable Father, pray for me (for me) to be saved from devilment.’
These examples illustrate two properties of OR non-finite subjects, which differ with respect to
the properties of NSs in infinitive clauses in PDR (cf. § 7.2.1), more specifically, the possibility
of having split antecedents (compare (19b) and (20a) below), and the availability of an NS
without a local antecedent (compare (19c) and (20b) below):
(20) a * [Ty so mnoj]i poobeščali [ei pojti vmeste v kino].
you with me promised go.INF together to cinema
b * Prosi [dlja menjai] [ei byt’ izbavlen ot takix neprijatnostej].
ask for me be.INF saved.NOM from such annoyances
Unlike in PDR, the properties of Obligatory Control described by Hornstein (1999) are absent
in this type of infinitive embedded structures. Split antecedents and the lack of a local
antecedent have been illustrated in (19b); Madariaga (2011) illustrates the lack of sloppy
interpretation under ellipsis, and the possibility of disjoint reference between an infinitive NS
and its potential antecedent in the matrix clause.22
21 These constructions are more frequent in Old Church Slavonic texts and in OR literary texts than in colloquial
texts (i.e. letters), which indicates that the mechanism of embedding events by means of an infinitive clause was
probably already in decline in spoken OR. In any case, overt dative subjects and their related properties survived
in such constructions for several centuries, as we will see in this section.
22 The same proscription on split antecedents (Livitz 2014: 74ff), as well as the other properties of OC, also
operates in the case of NSs of PDR finite control structures (Tsedryk 2012); cf. footnote 4.
22
The relevant data suggest that these types of embedded infinitive structures in early Slavic
lacked any type of syntactic control (although semantic control was present). Thus, these
clauses patterned with the general strategy of juxtaposition as a mechanism of semantic
embedding or subordination of events in other early Indo-European languages, which scholars
usually interpret as syntactically adjoined clauses (cf. fn. 19). Here are some arguments in
favour of this interpretation (for a more complete account of ‘No Control’ structures in OR, cf.
Madariaga 2011):
(i) An embedded infinitive clause could not only show disjoint time reference, but even display
the corresponding tense markers overtly, as shown in (21), where the matrix clause is specified
for the past tense, whereas the embedded infinitive clause has an overt future tense marker (this
co-occurrence of different overt tense markers is impossible in PDR):
(21) Neuspelosja [e budetъ čego nyne dostupiti], i to i vpered ne
was not time.PAST will be.FUT what now reach.INF also this also later not
ujdet. (OR: Ivan the Terrible’s Epistles 230)
will go.FUT
‘If you did not have time to conquer some (of these regions) immediately (in the future),
this (region) will not be yours later on.’
(ii) The complementary distribution of NSs with overt pronouns in OR embedded clauses is
characteristic of NOC structures (Hornstein 1999: 74), but there is a property which shows
rather a total lack of control in the OR structures under study, namely, the fact that NSs are also
in complementary distribution with overt NPs (cf. (19a) above). A similar argument comes
from the possibility of using a resumptive overt pronoun, that is, ‘repeating’ the antecedent in
the matrix clause as a dative subject within the embedded infinitive clause. In these cases, the
‘repeated’ embedded subject is interpreted as an emphatic pronoun; cf. the examples in (22):
(22a) is from the eleventh to twelfth centuries and (22b) from several centuries later.
(OR: Sinodal Patericon 63)
(22) a Starьcьi že reče [ne mošči emui trьpěti truda postьnaago].
old.NOM PRT said not be able.INF him.DAT suffer.INF task of fasting
‘The old man said that he was not able to stand the difficulties of fasting.’
b I oběščanьe ixъi vsěxъ bylo, [čto bylo imъ žiti vъ pravoslavnoj
and promise theirs all was that was them.DAT live.INF in orthodox
krestьjanskoj věrě krěpko, a kъ svoej prežnej musulmanskoj verě ne
Christian faith strongly and to their former Muslim faith not
obraščatisja]. (OR: Deed # 263, 1563 year)
23
return.INF
‘They all promised that they would live in the Christian orthodox faith, and that
they would not return to their former Muslim faith.’
According to Borkovskij (1979: 135) and Lomtev (1956), in late Middle Russian (from the
sixteenth century), these properties were lost. At this point, the new pattern of embedded non-
finite clauses described in § 7.2.1 arose:
(1) Infinitive embedded clauses, in particular, complements of volitional and jussive verbs, and
most purposes clauses with coreferent subjects were reduced to OC. Thus, they show the
properties described in § 7.2.1 (the unavailability of split antecedents, sloppy interpretation
under ellipsis, the presence of a local c-commanding antecedent, etc; cf. Madariaga 2011).
(2) Overt dative subjects, co-referent or not with respect to an element in the matrix clause,
were lost in embedded non-finite constructions, the complement of jussive and volitional verbs
(compare (23) to (19b) above):
(23) Myi poobeščali [(*nami / *imj) pojti vmeste v kino].
we promised us.DAT / them.DAT go.INF together to cinema
‘We promised to go to the cinema together’
(3) The ability to check the dative case in embedded infinitive contexts was lost together with
the loss of overt dative subjects (it was preserved in root infinitive clauses, as shown in § 7.3.1).
The reason will be explained in § 7.4, where, in a nutshell, I will argue that NSs in infinitive
embedded clauses became traces of a movement of the subject to get case in the matrix clause
(within a Movement Theory of Control), namely due to the loss of the ability of infinitives to
assign case in embedded contexts.23 In root contexts, however, the modal feature on C licenses
23 I will not take into account Landau’s (2008) argument that infinitive clauses (more specifically, a C head
associated to certain infinitive clauses) are able to assign the dative case in PDR under OC. In fact, PDR OC
infinitive clauses, strongly contrasting with previous stages of the language, lack the ability to license the dative
case (unlike in root infinitive clauses). In fact, Landau’s (2008) survey ignores the following considerations, crucial
to defend his account:
(i) Secondary non-verbal predicates (adjectives, nouns or participles) in infinitive clauses either show up in the
same case as their antecedent in the matrix clause or receive the instrumental case within a sort of Small Clause or
Predicative Phrase (Bailyn 1995); e.g. Vanja xočet [pojti na večerinku trezvyj / trezvym / odin] ‘Vania.NOM wants
to go to the party sober.NOM/INST / one.NOM (=alone)’. They never show up in the dative case in PDR control
structures, a fact overlooked by Landau (2008), whose survey illustrates only instances of Floating Quantifiers
(the so-called semi-predicatives). In other words, we never have the dative case on secondary predicates in the
same exact configurations as Landau’s (2008) examples of Floating Quantifiers (FQs) within infinitive OC clauses.
24
the dative case on the subject (and related predicates or floating quantifiers), following Bailyn
(2012). In the case of Floating Quantifiers in embedded infinitive clauses in PDR, unlike OR
(cf. example (19b)), in regular conditions, they show up with the nominative case, in the same
way as their antecedent in the matrix clause (24):
(24) Myi poobeščali [ei pojti oba / *oboim vmeste v kino].
we.NOM promised go.INF both.NOM/*DAT together to cinema
‘We promised to both go to the cinema together.’
(4) The availability of overt tense markers, with disjoint reference or not with regard to the
matrix tense, was lost in embedded non-finite constructions (compare (25) to (21) above):
(25) Raz tyi ne uspel [ei ob ètom (*budet / *bylo) podumat’], ja tebe ob’’jasnju.
once you not had time about this will / was think.INF I you explain
‘Once you did not have any time to think about it, I would explain it to you.
(5) Replacement of all other types of infinitive clauses by finite CPs headed by an overt C-
element (all complements of declarative and perceptive verbs, most complements of nominal
elements; and complements of volitional, jussive verbs, as well as purpose clauses with non-
This fact strongly contrasts with the situation in OR, where the dative case was regularly licensed in infinitive
embedded clauses both on subjects and related non-verbal predicates or FQs.
(ii) Secondary predicates in control structures obligatorily undergo the instrumental predicate case when case
transmission from the matrix antecedent is blocked (the cases described as instances of ‘case independence’ in
Landau 2008), e.g. Vanja perestal pit’, [čtoby [pojti na večerinku trezvym]] ‘Vania.NOM stopped drinking in order
go to the party sober.INST’. This gives the strong impression that inserting the secondary predicate in a Small
Clause, licensing its own particular case, functions as a kind of last resort operation to fulfil morphological case
realization whenever the other mechanism of case marking is not available.
(iii) For independent reasons, detailed in Madariaga (2006), quantifiers and FQs in Russian never undergo the
instrumental predicate case in the language (they do not form a Small Clause / Predicate Phrase).
(iv) And it is precisely the instances of Landau’s ‘case independence’, i.e. those in which case transmission into
control structures is blocked, and secondary predicates in them regularly undergo instrumental case marking, the
only ones in which FQs must show up in the dative case; e.g. Vanja porugalsja so vsemi, [čtoby [pojti na večerinku
odnomu]] ‘Vania.NOM got angry to everyone in order go to the party one.DAT (=alone)’.
The combination of these conditions gives us a very different picture of the situation of the dative case in control
infinitive structures in PDR: if case transmission from their antecedent is blocked, FQs, which for independent
reasons cannot be encoded with the instrumental predicate case, undergo a more radical last resort operation,
maybe a special morphological rule with no underlying syntactic motivation, as proposed for quirkies in
Sigurðsson (2012a). This last resort case marking is realized as dative, not surprisingly, in a sort of relictic
preservation of the old widespread dative infinitive case marking. For further criticism of Landau (2008)
concerning methodological considerations, and the data themselves, see footnote 38 in Tsedryk (2012).
25
co-referent subjects). Consider the examples in (26): (26a) is parallel to (19a) and illustrates an
infinitive clause complement of a perceptive verb, whereas (26b) is parallel to (22b), a
complement of a noun. None of these examples can be rendered with an infinitive clause
nowadays, but must be expressed with the help of a finite CP:
(26) a On uslyšal, [kak / čto zvučit kolokol / *zvučit’ kolokolu].
he heard how / that sounds.3.SG. bell.NOM / sound.INF bell.DAT
‘He heard that a bell was ringing.’
b Onii dali oběščan’e, [čto onii / ei budut žit’ / *im žit’ po pravilam].
and gave promise that they.NOM will.3.PL live them.DAT live by rules
‘They promised that they would live according to the rules.’
To summarize, early Slavic infinitive structures were close to juxtaposed constructions, in a
sort of semantic (not syntactic) embedding, strongly contrasting with modern embedded
infinitive constructions. From a syntactic point of view, OR embedded non-finite clauses could
not display any syntactic control, and could be characterized as adjoined clauses, like similar
‘subordinated’ structures, in particular, relative clauses, absolute constructions and infinitive
clauses in other early Indo-European languages (cf. fn. 17), while their PDR equivalents display
the properties of syntactic control in fully argumental configurations (complements of a verb).
7.3.3. Interim summary: the development of null subjects in Russian
The shift experienced by Russian from a consistent NS-language status into a partial NS-
language involved the following changes in root contexts: (i) loss of the rich tense morphology
of early Slavic in favour of a system based on aspectual distinctions (independent development
with respect to the loss of person morphology); (ii) reanalysis of emphatic strong personal
pronouns as weak pronouns (asymmetric development in 1st/2nd vs. 3rd person); and (iii) loss
of the auxiliary conveying person morphology only in the past tense. The period of loss of the
obligatory insertion of NSs in non-emphatic contexts can be posited between the fifteenth and
seventeenth centuries (for 1st/2nd person and 3rd person, respectively).
In § 7.2.1, on the one hand, I argued that the distribution of NSs in finite and non-finite contexts
in PDR is very similar. Moreover, in OR, overt NPs and overt or null pronouns had a similar
distribution in infinitive clauses (embedded or root) and in finite contexts. As shown in (19),
null (non-emphatic) pronouns, overt (emphatic) pronouns, and overt NPs could be used in
embedded non-finite contexts in OR, in the same way as in finite contexts (cf. § 7.3.1). On the
other hand, the OR dative non-finite embedded subjects and the properties associated with ‘no
control’ structures (the properties typical of modern NOC contexts plus the availability of overt
26
T markers with disjoint reference and ‘resumptive’ overt pronouns in embedded contexts), were
lost by the sixteenth century, being replaced by either OC infinitive structures or finite CPs,
depending on the specific configuration
7.4. The historical change and its implications for pro
In this section, I will propose a reason for the diachronic change of NSs in Russian, giving a
formal explanation for the relevant facts described in §§ 7.2 and 7.3.
First, I will introduce two concepts that will aid an understanding of the change under study:
(i) Movement Theory of Control (Hornstein 1999; Boeckx and Hornstein 2007; Boeckx et al.
2010a): embedded NSs in OC structures are traces or copies, resulting from the movement of
an embedded subject into a higher position in the matrix clause.
(ii) Every other referential NS is pro (or some version of it), i.e. deleted or elided real pronouns,
with some featural content.
Taking these two ideas into account, let us review the properties of NSs in the crucial periods
of the Russian language, as well as the change between them:
(1) In Old Russian:
(i) OR infinitive structures of semantic subordination, which we would consider as
‘embedded’ nowadays, did not display any syntactic control. This was probably
a vestige of the early Indo-European character of ‘subordinate’ clauses as
adjoined and not argumental (cf. § 7.3.2).
(ii) These structures could include non-emphatic NSs, emphatic overt dative
pronouns or overt dative NP subjects. The dative case was licensed by the
infinitive itself, as it is nowadays in PDR root infinitive clauses (cf. Franks 1995;
Bailyn 2012). In the event that a floating quantifier or a secondary non-verbal
predicate was used, it also showed up in the dative case.
(iii)The distribution of subjects in embedded contexts mirrored that of root finite
subjects: non-emphatic NSs, emphatic overt pronouns, and regular NP subjects.
In the event of finite contexts, the case valued on these subjects was nominative.
(2) In later Russian:
(i) PDR non-finite structures in the complement position of volitional and jussive verbs
can only be instances of OC, where the interpretative reference of the embedded subject
correlates with the corresponding syntactic properties (locally controlling antecedent,
etc.). The associated secondary predicates are marked with the instrumental predicate
case, or ‘agree’ with their antecedent in the matrix clause. The only exceptions are
27
floating quantifiers which, for independent reasons, are not licensed in an instrumental
case configuration and are therefore marked by case transmission. When case
transmission is blocked, they undergo a special morphological rule, a relictive use of
the old dative case associated with non-finite structures in OR (see footnote 22).
(ii) According to the Movement Theory of Control, in such constructions the embedded
subject raises into the matrix clause to get (nominative) case licensed; the nominative
case is also reflected on any FQs that can be used in the embedded infinitive clause.
(iii) Following Tsedryk (2013), NSs in embedded finite clauses are also instances of OC
(movement of the embedded subject into a matrix subject position). In this case, I
propose that the subject raises for discourse reasons: NSs in PDR are bound topics; if
embedded contexts lack a topic position, the subject raises into the matrix clause to a
position where it can be interpreted as a topic.24
24 NSs in embedded finite clauses are licensed only in what Tsedryk (2013) describes as a ‘nominative chain’,
formed by movement from an A-position into the same A-position without anything intervening or blocking the
chain. He argues that there is nothing in UG blocking internal merge of nominative subjects, and shows that the
factors that can block this type of movement do not apply in this specific configuration in Russian. In fact, the
that-trace effect usually operating in čto-clauses in PDR would have no effect in the case of movement of a
nominative subject to a higher nominative subject position. I will not address here the technical details, which can
be found in Tsedryk (2012, 2013): in a nutshell, he argues that C in Russian has a weak edge feature, i.e. it is not
a strong phase in čto-clauses and therefore, it is opaque for A′-movement, yielding the that-trace effect shown in
(i), but it is transparent for A-movement, allowing nominative chains involving NSs in embedded finite clauses,
as in (ii). As expected, if we try to perform an A′-movement of the subject in (ii), in the same way as in (i), the
resulting structure is also ungrammatical (iii):
(i) * Ktoi ty skazal, čto ei/j prigotovit pirog? (Tsedryk 2012)
who.NOM you said that will cook pie
‘Who did you say that he/she would prepare a pie?’
(ii) Lenai skazala, čto ei/*j prigotovit pirog.
Lena.NOM said that will cook pie
‘Lena said that she would prepare a pie.’
(iii) *?? Lenai (,) ty skazal, čto ei/j prigotovit pirog.
Lena.NOM you said that will cook pie
‘Lena, you said that she would prepare a pie.’
The that-trace effect is also not an argument against the fact that the moved subject checks some topic feature in
the matrix clause. The simplest reason is that the moved subject does not need to raise further into the matrix CP
to check the topic feature, but can check it from the matrix subject position. However, even if we propose a further
movement into the matrix A′-position, this would be just a short A′-movement of an element already present in
28
The features of the two patterns in the relevant periods are summarized in Table 1:
Feature Early Slavic PDR
Obligatory NS in
non-emphatic
contexts
In root clauses
√ *
In embedded clauses
√ *
Overt tense
markers available
In root non-finite
clauses √ √
In embedded non-finite
clauses √ *
Overt dative
subjects available
In root non-finite
clauses √ √
In embedded non-finite
clauses √ *
Disjoint reference
of subjects
available
In finite embedded
clauses √ √
In non-finite embedded
clauses √ *
Obligatory
controlled NS
In finite embedded
clauses
* √
In non-finite embedded
clauses
* √
Table 1: Compared features in early Slavic and PDR
As we infer from Table 7.1, there are two main asymmetries in the distribution of the
relevant features: (i) root versus embedded clauses: root clauses do not basically change through
the history of Russian (except for the distribution of NSs in them), while embedded clauses
change with respect to all the features introduced here; and (ii) ‘obligatory’ versus ‘optional’
factors: both root NSs and embedded NSs shifted in Middle Russian to the reverse of the status
they used to have: obligatory root NSs became optional in non-emphatic contexts, while
embedded NSs became obligatorily controlled in both non-finite and finite clauses.
As for infinitive embedded clauses, the change described can be represented as the shift
from Grammar 1 into Grammar 2 in (27) below. This change took place in the transition
the matrix clause, thereby circumventing the that-trace effect. If so, we would expect such a movement to also be
licit in the case of wh-questions and, actually, it is:
(iv) Ktoi skazal, čto ei/*j prigotovit pirog? (From Tsedryk 2012)
who.NOM said that will cook pie
‘Who said that he/she would prepare a pie?’
29
between the sixteenth and seventeenth centuries, roughly together with the loss of the consistent
NS-language character of Russian, that is, the loss of obligatory pro in non-emphatic contexts:
(27) Grammar 1: [ NPi matrix-V [ (pro / pronoun / NP)i/j V.INF ]] >
Grammar 2: [ NPi matrix-V [ ti V.INF ]
In order to account for (27), we could resort to the change in some relevant feature of infinitive
T. This is the way in which synchronic studies account for the asymmetry between root and
embedded infinitive clauses in PDR:
(i) Franks (1995) attributes the ability of Russian root infinitives to license case or overt
subjects to a [+T] feature (cf. (16b)); embedded infinitive clauses would lack such a
feature, thus, they do not license overt dative subjects in PDR.
(ii) More recent accounts relate that ability to value case or license overt subjects to the
modal value always associated with Russian root infinitives, but absent in embedded
infinitives, as in Bailyn (2012: 169ff), who puts the ability to assign the dative case in a
[+modal] feature on C in root, but not embedded, infinitive clauses.
Let us then assume that the difference between early Slavic and PDR is that the relevant feature,
which distinguishes root and embedded infinitive clauses nowadays, was the same in all
infinitival forms in OR. Assuming the Inertia of syntax in diachronic change (Longobardi 2001),
this unique feature of OR infinitive structures could not split spontaneously and become
differentiated in root versus embedded clauses for no reason. Learners had to detect some new
cue in their PLD, which led them to reanalyse the relevant feature in the case of embedded
clauses, but not in root clauses.
In other words, looking at the diachronic facts presented here, a question remains
unexplained: why should the relevant feature be lost only in embedded contexts? From an
acquisitional point of view, learners had positive evidence to keep acquiring that feature,
because of the presence of overt dative pronouns and NPs in root infinitive environments. In
other words, the loss of overt dative subjects in embedded infinitive clauses could not be due
to a spontaneous change in the ability of Russian infinitives to license case in general terms, as
root infinitives licensed (and still license) the dative case on their subjects with almost no
change through the history of Russian. Thus, a different and independent cue had to give rise
to the split between root and embedded infinitive clauses, as I will show in what follows.
A good starting point to explain the facts can be found in certain synchronic accounts which
attribute the contrast between PDR root and embedded clauses to different types of subjects,
corresponding to the different features conveyed by the infinitive in each syntactic environment.
Thus, Franks and Hornstein (1992) identify (non-case-licensers) embedded infinitive subjects
30
with instances of anaphoric pro, while root infinitive clauses, licensing case, would include a
pronominal pro. This is a pre-version of the Movement Theory of Control adopted in this
chapter.
If this is true, and there is a link between the nature of a non-finite subject and the properties
of the corresponding infinitival, a possible explanation for (27) emerges: for some independent
reason, the properties of subjects could have changed in embedded infinitive clauses, but not in
root infinitive clauses, producing an impact on the properties of the infinitive clauses
themselves. This initial independent trigger could be, in fact, the change experienced by NSs,
as outlined in § 7.3:
(i) Initially, OR was a consistent NS-language, in which pro or deleted pronouns were
compulsory in every non-emphatic context.
(ii) When Russian developed overt weak pronouns, pro or deleted pronouns stopped being
the null element obligatorily parsed in non-emphatic contexts (both in finite and non-
finite contexts).
(iii)Therefore, learners no longer needed to parse pro in the gap they encountered in the
subject position of infinitive embedded constructions.
(iv) According to the principle ‘movement is preferred over pronominalization’ (proposed
by Boeckx et al. 2010a, Boeckx and Hornstein 2007),25 learners started to parse the
embedded non-finite gap as an NP-trace or copy, resulting from internal merge of the
embedded subject in the matrix clause.
(v) OC structures arose in embedded clauses (both finite and non-finite), with the
corresponding properties, reviewed in § 7.3.2.
Non-finite embedded subjects raise to the matrix clause in order to get case, whereas finite
embedded subjects move in order to get to a topic position. The distinct motivations for such
movement have further implications:
(1) As far as the feature in C motivating the existence of a topic is optional (determined by
discourse factors), embedded finite subjects can be null or overt, that is they can raise to
the matrix clause in order to satisfy a topic feature, when it is present, or just stay at the
embedded [Sp, TP] position, when that topic feature is absent (in ‘neutral’ discourse
conditions). In this sense, they show the same syntactic behaviour as root subjects in
25 Somewhat in line with Chomsky’s (2013: 41) observation that there is no basis for believing that External Merge
is simpler than Internal Merge, when he says that ‘If anything, IM is simpler, since it requires vastly less search
than EM (which must access the workspace of already generated objects and the lexicon).’
31
PDR; that is, NSs undergo specific restrictions and, even in the cases when an NS is
licensed, the corresponding overt equivalent is also licit.26
(2) In embedded infinitive clauses, however, there is no other option than having an NS in
PDR, contrasting with the existence of overt infinitive subjects in OR. The reason is that
infinitive subjects, after the reanalysis of non-finite structures such as OC structures,
cannot receive case within the embedded clause and therefore must obligatorily raise to
a case position in the matrix clause.
7.5. Concluding remarks
In this final section, I will sketch the implications that follow from the analysis presented in this
chapter: first of all, partial NS-languages do not seem to be as homogeneous as we would expect
if they represent a particular setting of a consistent parameter. Comparing the properties of NSs
in PDR and other partial NS-languages, we observe that there are differential features which,
in the case of Russian, have been clearly determined by the historical development of other
(more or less independent) parts of the grammar. In general, pro-drop does not seem to be a
homogeneous parameter, but rather the conspiracy of a bundle of properties (such as verb
movement and personal morphology), which sometimes interact with completely independent
phenomena such as control structures and other null elements.
The diachronic development of the Russian language determined the common properties
that NSs in finite and non-finite constructions display nowadays, rendering in some way a
seemingly more unified pattern of pro-drop than in other partial NS-languages. In particular,
the reanalysis of the subject gap in embedded contexts (the replacement of an obligatory
pronominalization by a null element resulting from movement), as expected, affected all
embedded contexts, both finite and non-finite.
A further implication of the account of syntactic change presented here is the following:
the relevant gap in the structures under study, that is the null element corresponding to a
referential NS, contains a real pronominalization of some sort. First, we have established the
fact that rich agreement was an independent phenomenon in the change process reviewed in
26 Interestingly, a drastic drop in frequency of 3rd person NSs took place in embedded finite contexts earlier than
in root clauses; roughly at the beginning of the sixteenth century, overt weak pronouns experienced a significant
increase in embedded clauses, a century earlier than in root clauses, as shown by Luraghi and Pinelli (2015). This
observation is fully consistent with the proposal on NSs in embedded finite clauses presented in this chapter;
namely, that the lack of an appropriate topic position in embedded contexts triggered in them a faster development
(a higher number) of weak overt pronouns during this transition stage of the language.
32
this chapter, both historically and synchronically. Impoverishment of verbal person
morphology could not give rise to change because (i) it cannot account for the change in non-
finite clauses, which always lacked personal morphology, (ii) person morphology on the verb
was preserved in the course of the history of Russian (except for the past forms), and (iii) even
in the case of loss of personal morphology in the past tense, it developed in a chronologically
independent way with respect to the change analysed in this chapter. Second, the reanalysis
presented here suggests that this null element must be ‘something’ parsable in the subject
position; otherwise, it would not be able to enter into competition for that position with other
alternative null elements. In the case under study, the alternative to a null embedded pronoun
was a trace or copy, which precisely occupies the required position, as it is the result of the
movement of an embedded subject into a matrix clause.
Finally, this case study suggests that pro (or a version of it) is the weak pronoun in
consistent pro-drop languages. In this sense, the Avoid Pronoun principle (Chomsky 1981;
Rizzi 1997a), at least from a diachronic point of view, can be reformulated from the point of
view of the (non-)obligatory character of pro, instead of its availability in a specific syntactic
position. In simpler words, diachronically speaking, the pertinent question will not be whether
pro is available or not in a language, but whether pro is obligatory or not in a relevant context.
Why? On the one hand, the ability of infinitives to license case never changed in root clauses;
therefore, from the point of view of learners, embedded infinitives could not lose their ability
to license case spontaneously. On the other hand, null pronouns were always available in
Russian, so we cannot attribute the changes in embedded infinitive clauses to a general loss of
the availability of NSs or pro in the language. Instead, the loss of the obligatory realization of
pro in non-emphatic contexts could lead in a straightforward way to a crucial effect in learners
that was necessary for change to take place: when learners stopped being compelled to interpret
every non-emphatic subject position in root contexts as a null pronoun, they started to not parse
embedded gaps as null pronouns either. At this point, reanalysis took place: according to the
preference for movement over pronominalization, learners started to interpret the null category
of embedded clauses as a trace or copy instead.
References
References to old texts
33
Deeds: Akademija nauk (1836). Akty, sobrannye v bibliotekax i arxivax Rossijskoj Imperii
Arxeografičeskoju èkspedicieju imperatorskoj Akademii nauk -tom I. Sankt-Peterburg.
Ivan the Terrible’s letters: Lixačev, D. S., Ja. S. Lur’e, and V. P. Adrianova-Peretc (eds.)
(1951). ‘Poslanie Ivana Groznogo pol’skomu korolju Stefanu Batoriju 1581’. Poslanija
Ivana Groznogo. Moskva – Leningrad.
Laurentian Chronicle: Karskij, Je. F. (ed.) (2001 [11926-28]). Lavrent’evskaja letopis’ –
Suzdal’skaja letopis’. Polnoe sobranie russkix letopisej - tom I. Moskva.
Mstislav’s letter: Obnorskij, S. P., and S. G. Barxudarov (eds.) (1952). ‘Gramota velikogo
knjazja Mstislava Vladimiroviča i syny ego Vsevoloda novgorodskomu Jur’jevu
monastyrju’. Xrestomatii po isstorii russkogo jazyka. Moskva.
1st Novgorod Chronicle: Nasonov, A. N. (ed.) (2000 [11950]). Novgorodskaja Pervaja Letopis’
staršego i mladšego izvodov. Polnoe sobranie russkix letopisej – tom III. Moskva,
Novgorod letters: Zaliznjak, A. A. (2004) Drevnenovgorodskij dialect. Moskva.
Sermon on Law and Grace: V. Sreznevskij (ed.) (1893). ‘Slovo Ilariona o Zakone i Blagodati’.
Musin-Puškinskij sbornik 1414. Sankt Peterburg.
Sinodal Patericon: Golyščenko, V. S., and V. F. Dubrovina (eds.) (1967). Sinajskij Paterik.
Moskva.
Suzdal Chronicle: Karskij, Je. F. (ed.) (2001 [11926-28]). Lavrent’evskaja letopis’ –
Suzdal’skaja letopis’. Polnoe sobranie russkix letopisej - tom I. Moskva.
Other references
Alexiadou, Artemis, and Elena Anagnostopoulou (1998). ‘Parametrizing AGR: Word order,
V-movement, and EPP-checking’, Natural Language and Linguistic Theory 16: 491-540.
Andersen, Henning (1970). ‘The dative of subordination in Baltic and Slavic’, in Thomas
Magner and William Schmalstieg (eds.), Baltic Linguistics. University Park: Pennsylvania
State University Press, 1-9.
Avanesov R. I., and Ivanov, V. V. (1982). Istoriceskaja grammatika russkogo jazyka. Moskva:
Nauka.
Barbosa, Pilar (1995). Null subjects. Ph.D. Dissertation MIT. Cambridge, Mass.
Bar-Shalom, Eva, and William Snyder (1997). ‘Optional Infinitives in Russian and their
Implications for the Pro-Drop Debate’, Formal Approaches to Slavic Linguistics 5: the
Indiana Meeting, Michigan: Michigan Slavic Publications, 38-47.
34
Bailyn, John F. (1995). A Configurational Approach to Russian 'Free' Word Order. Doctoral
Dissertation, Cornell University.
Bailyn, John F. (2012). The Syntax of Russian. Cambridge: Cambridge University Press.
Biberauer, Theresa, Anders Holmberg, Ian Roberts and Michelle Sheehan (eds.) (2010).
Parametric variation. Cambridge: Cambridge University Press.
Borer, Hagit (1986). ‘I-subjects’, Linguistic Inquiry 17: 375-416.
Boeckx, Cedric, Norbert Hornstein, and Jairo Nunes (2010). Control as Movement. Cambridge:
Cambridge University Press.
Boeckx, Cedric, and Norbert Hornstein (2007). ‘On (non-)obligatory control’, in W. D. Davis
and S. Dubinsky (eds.), New Horizons in the Analysis of Control and Raising. Springer,
251-262.
Borkovskij, V. I. (1949). Sintaksis drevnerusskix gramot. Prostoje predloženie. L’vov:
L’vovskij Gosudarstvennyj Institut.
Borkovskij, V. I. (ed.) (1968). Sravnitel´no-istoričeskij sintaksis vostočnoslavjanskich jazykov.
Tipy prostogo predloženija. Moskva: Nauka.
Borkovskij, V. I. (1978). Istoričeskaja grammatika russkogo jazyka: sintaksis - prostoe
predloženie. Moskva: Nauka.
Borkovskij, V. I. (1979). Istoričeskaja grammatika russkogo jazyka: sintaksis - složnoje
predloženie. Moskva: Nauka.
Borkovskij, V. I. & P. S. Kuznecov. (1963). Istoričeskaja grammatika russkogo jazyka.
Moskva: AN SSSR.
Cardinaletti, A. (1997). ‘Subjects and Clause Structure’, in Liliane Haegeman (ed.), The new
comparative syntax. London: Longman, 33-64.
Černyx, P. Ja. (1954). Istoričeskaja grammatika russkogo jazyka. Moskva: Nauka.
Chomsky, Noam (1981). Lectures on government and binding. Dordrecht: Foris.
Chomsky, Noam. (2013). Problems of projection. Lingua 130: 33–49.
Comrie, Bernard (1974). ‘The Second Dative: a Transformational Approach’, in Richard Brecht
and Catherine Chvany. Slavic Transformational Syntax. Ann Arbor: University of
Michigan, 123-150.
Demjjanow, Assinja, and Anatoli Strigin (2000). ‘Case Assignment to Conceptual Structures:
The Russian Instrumental’, in Marcus Kracht and Anatoli Strigin (eds.), Papers on the
Interpretation of Case. Potsdam: Universität Potsdam, 75–107.
Disterheft, Dorothy (1980). The syntactic development of the infinitive in Indo-European.
Columbus: Slavica Publishers.
35
Disterheft, Dorothy (1997). ‘The evolution of Indo-European infinitives’, in Dorothy
Disterheft, Martin Hud and John Greppin (eds.), Ancient languages and philology.
Studies in Honor of Jaan Puhvel, volume 1. Washington: Institute for the Study of Man,
101-122.
Fortuin, Egbert (2005). ‘From possibility to necessity: the semantic spectrum of the dative-
infinitive construction in Russian’, in: Hansen, B. & Karlík, P. (eds.), Modality in
Slavonic Languages. New Perspectives, München: Otto Sagner, 39-60.
Franks, Steven (1995). Parameters of Slavic Morphosyntax. Oxford University Press.
Franks, Steven, and Norbert Hornstein (1992). ‘Second Predication in Russian and Proper
Government of PRO’, in Richard Larson, Sabine Iatridou, Utpal Lahiri and James
Higginbothan (eds.), Control and Grammar. Dordrecht: Kluwer Academia Publishers, 1-
50.
Frascarelli, Mara (2007). ‘Subjects, topics and the interpretation of referential pro’, Natural
Language and Linguistic Theory 25: 691–734.
Frascarelli, Mara (2014). ‘Avoid pronoun in consistent and partial pro-drop languages: a
comparative interface analysis’, Paper presented at the workshop Understanding pro-
drop: a synchronic and diachronic perspective (Trento, 19-21 June 2014).
Frascarelli, Mara, and Roland Hinterhoelzl (2007). ‘Types of Topics in German and Italian’, in
S. Winkler and K. Schwabe (eds.), On Information Structure, Meaning and Form.
Amsterdam/Philadelphia: John Benjamins, 87-116.
Garrett, Andrew (1994). ‘Relative clause syntax in Lycian and Hittite’, Die Sprache 36. 29-69.
Holmberg, Anders (2005). ‘Is There a Little Pro? Evidence from Finnish’, Linguistic Inquiry
36: 533-564.
Holmberg, Anders, A. Nayudu, and M. Sheehan (2009). ‘Three partial null-subject languages:
a comparison of Brazilian Portuguese, Finnish and Marathi’, Studia Linguistica 63: 59-
97.
Hornstein, Norbert (1999). ‘On Control’, Linguistic Inquiry 30/1: 69-96.
Istrina, Je. S. (1923). Sintaksičeskie javlenija sinodal’nogo spiska pervoj novgorodskoj letopisi.
Sankt-Peterburg: RAN.
Ivanov, V. V. (1990). Istoričeskaja grammatika russkogo jazyka. Moskva: Prosveščenie.
Jung, Hakyung (2014). ‘The syntax of the be-auxiliary and D-feature lowering in Old North
Russian’, Ms. Seoul National University.
Keydana, Götz (2013). ‘Infinitive im R̥gveda: Formen, Funktion, Diachronie’, Leiden: Brill.
36
Kiparsky, Paul (1995). ‘Indo-European origins of Germanic syntax’, in Adrian Battye and Ian
Roberts (eds.), Clause structure and language change. Oxford: Oxford University Press.
Kotin, M. L. (2012). ‘Impersonal dative constructions as covert root-modality patterns’, in
Werner Abraham & Elisabeth Leiss (eds.) Covert patterns of modality, Cambridge:
Cambridge Scholars Publishing, 153-174.
Landau, Idan (2004). ‘The scale of finiteness and the calculus of control’, Natural Language
and Linguistic Theory 22: 811-877.
Landau, Idan (2008). ‘Two Routes of Control: Evidence from Case Transmission in Russian’,
Natural Language and Linguistic Theory 26/4: 877-924
Lightfoot, David (1991). How to Set Parameters: Arguments from Language Change.
Cambridge, MA: MIT Press.
Lightfoot, David (1999). The Development of Language: Acquisition, change and evolution.
Blackwell.
Lindseth, Martina (1998). Null-subject properties of Slavic languages: with special reference
to Russian, Czech and Sorbian. Otto Sagner.
Livitz, Inna (2014). Deriving silence through internal reference: focus on pronouns. Ph.D.
Dissertation New York University.
Lomtev, T. P. (1956). Očerki po istoričeskomu sintaksisu russkogo jazyka. Moskva:
Izdatel’stvo Moskovskogo Universiteta.
Longobardi, Giuseppe (2001). ‘Formal syntax, diachronic Minimalism, and etymology: The
history of French chez’, Linguistic Inquiry 32: 275-302.
Luraghi, Silvia, and Erica Pinelli (2015). ‘The loss of referential null subjects in Russian: what
subordinate clauses can tell us’, Presentation at the conference Slavic Corpus Linguistics:
the historical dimension. Tromsø, 21-22 April 2015.
Madariaga, Nerea (2006). ‘Why Semi-Predicative Items always Agree’, Journal of Slavic
Linguistics 14/1: 45-78.
Madariaga, Nerea (2011). ‘Infinitive clauses and dative subjects in Russian’, Russian
Linguistics 35: 301-29.
Madariaga, Nerea (2014). ‘On the early structure of Indo-European infinitive constructions:
Evidence from early Slavic’, Ms. University of the Basque Country.
Madariaga, Nerea (2015). ‘The decline of non-finiteness as a mechanism for embedding in East
Slavic’, Journal of Historical Linguistics 5/1: 139-174.
37
McShane, Marjorie (2009). ‘Subject ellipsis in Russian and Polish’, Studia Linguistica 63/1:
98-132.
Meyer, Roland (2009). ‘Zur Geschichte des referentiallen Nullsubjekts im Russischen’,
Zeitschrift für Slawistik 54: 375-97.
Meyer, Roland (2011). The history of Null Subjects in East Slavonic. A corpus based diachronic
investigation. Habilitation Thesis, University of Regensburg.
Müller, Gereon (2006). ‘Pro-Drop and Impoverishment’, in Patrick Brandt and Eric Fuß (eds.),
Form, Structure, and Grammar. A Festschrift Presented to Günther Grewendorf on
Occasion of his 60th Birthday. Berlin: Akademie Verlag, 93–115.
Nunes, Jairo (2010). ‘A Note on Wh-islands and Finite Control in Brazilian Portuguese’,
Estudos da Língua 8/2: 79-103.
Perlmutter, David, and John Moore (2002). ‘Language-Internal Explanation: The Distribution
of Russian Impersonals’, Language 78: 619–650.
Potebnja, A. A. (1958 [1888]). Iz zapisej po russkoj grammatike (I-II). Moskva:
Gosudartsvennoje učebno-pedagogičeskoje izdatel’stvo.
Rizzi, Luigi (1986). ‘Null objects in Italian and the theory of pro’, Linguistic Inquiry 17: 501-
57.
Rizzi, Luigi (1997). ‘A parametric approach to comparative syntax properties of the pronominal
system’, in Lilian Haegeman (ed.), The new comparative syntax. London: Longman,
268–285.
Roberts, Ian, and Anders Holmberg (2010). ‘Introduction: parameters in minimalist theory’, in
Theresa Biberauer, Anders Holmberg, Ian Roberts and Michelle Sheehan (eds.),
Parametric variation: null subjects in minimalist theory. Cambridge: Cambridge
University Press, 1-57.
Roberts, Ian (2010). ‘A deletion analysis of null subjects’, in Theresa Biberauer, Anders
Holmberg, Ian Roberts, and Michelle Sheehan (eds.), Parametric variation: null
subjects in minimalist theory. Cambridge: Cambridge University Press, 58-87.
Sigurðsson, Halldor (2011). ‘Conditions on argument drop’, Linguistic Inquiry 42: 267-304.
Sigurðsson, Halldor (2012). ‘Case variation: viruses and star wars’, Nordic Journal of
Linguistics 35/3: 313-342.
Tsedryk, Egor (2012). ‘Finite control in Russian as movement: subject extraction, feature
deficiency and parametric variation’, Ms., Saint Mary’s University.
38
Tsedryk, Egor (2013). ‘Internal Merge of nominative subjects and pro-drop in Russian’, in Shan
Luo (ed.), Proceedings of the 2013 Canadian Linguistics Association conference.
University of Victoria. http://homes.chass.utoronto.ca/~cla-acl/actes2013/actes2013.html
Wiemer, Björn (2017). ‘Main clause infinitival predicates and their equivalents in Slavic’, in
Lukasz Jedrzejowski and Ulrike Demske (eds.), Infinitives at the Syntax-Semantics
Interface: A Diachronic Perspective, Berlin: de Gruyter.
Zaliznjak, A. A. (2004). Drevnenovgorodskij dialect. Moskva: Jazyki slavjanskoj kul’tury.
Zaliznjak, A.A. (2008). Drevnerusskie ènklitiki. Moskva: Jazyki slavjanskoj kul’tury.