Subjecthood cannot be squishy: A case study of Russian dative subject constructions
|NP – noun phrase
IO – indirect object
NP1 – first argument of a verbal predicate in argument structure
Russian dative subject constructions:
DSC – dative subject construction
DBI – free infinitive construction
NOM – nominative
PREP – prepositional
REFL – reflexive
1SG – 1st person singular
While there may be some considerable or minor inter-language parallels, it is clear that subjects do not always behave the same cross-linguistically in terms of the subject properties they exhibit. Therefore, we must consider what it is to be a subject, for example, in the Russian language. A universal definition can only prove to be imprecise, while a more focussed, language-specific definition has the potential to aid advancement in the field by encouraging deeper linguistic analysis into the reasons why a nominal that has been labelled a subject does not behave like other subject-labelled nominals.
At first glance, constructions like the following both appear to exhibit so-called logical dative subjects:
- Tamare pozvonit’ mame.
tamara.dat call.inf mum.dat
‘Tamara has to call her mum.’
- Tamare ne sidit’-sja.
tamara.dat neg sit.3sg-refl
‘Tamara can’t seem to stay sitting.’
However, if we compare the argument structure of constructions appearing with a logical subject marked in the dative case in Russian, two groups emerge: those where the dative nominal appears in the first argument position of a verbal predicate, and those that do not. In this dissertation, I will illustrate that only the former group (consisting of the free-infinitive construction, constructions of potential and constructions of anteriority) can be considered dative subject constructions. Consequently, we must rethink the canon of what are often considered dative subject constructions in Russian. The dative experiencer, dative modal and semi-passive constructions must be struck from the list of dative subject constructions, since the subject in such utterances is in fact unrealised pro, as supported by the 3SG neuter verb agreement in such constructions. Having pinpointed the constructions that do feature true dative subjects, we can begin to create a more focussed definition and description of Russian dative subjects.
Until now, a nominal’s subject status has largely depended on its adherence to or deviation from the so-called ‘subject properties’, first devised by Keenan (1970). These nominative-centric properties have too often been used as a stand-in definition for subjecthood This approach, however, can only be described as ‘squishy’ – i.e. highly subjective and able to be shaped according to the analyst’s desire. Only a non-subjective definition of subjecthood can better serve to encourage deeper analysis of subject behaviour too often considered ‘deviant’.
While conducting their research, Barddal and Eythorsson (2005; to appear) realised that when conducting the so-called ‘subject tests’, it is always the first argument of the argument structure that is targeted. A definition of subjecthood in terms of argument structure forces the analyst to move away from viewing subjecthood as a gradient concept and fosters a more objective and principled approach to the definition of a grammatical subject. We will also assume a definition in terms of argument structure, more specifically that a subject is the first argument of a verbal predicate in surface structure within a sentence with neutral word order. By targeting analysis on the nominals found in the first argument position of a verbal predicate, language-specific definitions of subjects can start to be delineated. The analysis in this dissertation led to the conclusion that Russian dative subjects (i.e. dative-marked nominals found in the first argument position of the verbal predicate) are thematic experiencers of modality. This seems to be linked to the semantic – rather than grammatical – function of the dative case in non-finite contexts, as well as the cross-linguistic phenomenon whereby possessive constructions develop into modal ones.
Moore and Perlmutter (2000: 382) assert that ‘the term dative subject has come to be applied quite loosely to dative-marked nominals that display any subject behaviour whatsoever’. Yet, the term ‘dative subject’ is not applied uniformly. Indeed, there is an unnecessary plethora of terms floating around to label and classify non-nominative subjects. When discussing non-nominative subjects, the literature refers to oblique subjects; semi-, pseudo- or quasi-subjects; logical subjects and subject-like non-nominatives. These labels are often used to describe oblique, ‘subject-like’ nominals that only adhere to some of Keenan’s subject properties. If they do not fit so comprehensively to Keenan’s list of behaviours, they are immediately said to not possess ‘full’ subjecthood.
In his pioneering article on the matter of subjecthood, Keenan (1976) describes subjecthood as ‘a matter of degree’ and introduces a list of ‘universal subject properties’ which have since been widely used to assess a putative subject’s ‘subjectivity’. This includes properties such as reflexivisation, raising, control and conjunction reduction, among others. However, Keenan’s concept of a universal subject in terms of its subject properties is inherently flawed and cannot be maintained for two main reasons: Firstly, typological differences between languages render some of the tests inapplicable in certain languages. Secondly, Keenan’s definition of a subject in terms of its properties can only be described as ‘squishy’ – that is, ‘malleable’ in that it can be subjectively shaped according to the desire or whim of the analyst.
A simple example of a subject test not pertinent to all languages is syntactic position. While in languages such as Icelandic the syntactic position of a nominal can be employed to signify subject status, in others, such as German and Russian, syntactic position is merely used for emphasis. Russian utilises deviation from its neutral word order (SVO) to put emphasis on the rheme, or new information, which is placed the end of the sentence. So, while ‘Tamara’ (the subject) is fronted in utterance (3) with neutral word order, the emphasis lies on ‘jabloko’ (the object) in utterance (4) – but, importantly, it does not change the fundamental meaning of the statement. The apple is still being eaten by Tamara.
(3) Tamara est’ jabloko.
tamara.nom eat.3sg apple.acc
‘Tamara is eating the apple.’
(4) Jabloko est’ Tamara.
apple.acc eat.3sg tamara.nom
‘Tamara is eating the apple.’
Additionally, there is no set number of tests a nominal must pass in order to be considered a ‘full’ syntactic subject. Depending on how many tests it passes, the analyst rates the subject on their own personal scale. This can only be described as inadequate and highly subjective. Barddal & Eythorsson (to appear: 9) express this well when they correctly conclude that: ‘the subject properties themselves function as a stand-in definition of a subject [… and enforce] a gradient concept with major analytical problems’. The list of subject properties cannot serve as a definition in itself as it fosters a subjective – rather than an objective and principled – methodology on the part of the analyst. The properties are better referred to not as subject ‘tests’, but rather as ‘possible subject properties or behaviours’.
When conducting their research, Barddal & Eythorsson (to appear: 5) realised that conducting these so-called subjecthood ‘tests’ always involves targeting the first argument within the argument structure. This led Barddal & Eythorsson to assume the ‘working definition’ of a subject as the ‘first argument’ of a verb. Others have also defined the subject in terms of argument structure: Grillborzer (2011) has defined it the ‘leftmost argument’ (more explicitly as ‘das Argument, das sich links vom Prädikat befindet, und zwar im unmarkierten Satz’, ‘the argument found to the left of the predicate, namely in an unmarked sentence’). A definition of subjecthood in terms of a nominal’s relation to argument structure and not its behavioural properties has clear benefits. A definition is not gradient in nature, it is black and white. A NP can either be the first/leftmost argument, or it cannot. Only by having such a binary option is the analyst encouraged to not simply label a NP as a ‘part-subject’. This in turn sparks questions about the behaviours of such nominals and the reasons why the first arguments of verbal predicates behave the way they do, whether disparately or similarly to each other, both intra- and inter-linguistically.
What are the implications of this? Firstly, it is clear that we must abandon Keenan’s squishy list of subject properties and instead provide a clear, concrete, non-nominative-centric definition for subjecthood. This is extremely pertinent to dative-marked nominals in Russian, since they in particular contradict this concept of universal subject properties. Russian dative nominals are found in the first argument position of a verbal predicate and fulfil some of the behavioural subject criteria – but, crucially, not all. Only nominals in certain constructions with dative-marked nominals will, for example, participate in raising – a typical subject ‘test’. Dative-marked nominals in dative experiencer constructions, for instance, do not participate, as can been seen in (5):
(5a) Mame trudno odet’ rebjonka.
mum.dat hard dress child.acc
‘It is hard for the mother to dress the child.’
(5b) *Rebjonku trudno byt’ odetym mamoj.
child.dat hard be.inf dressed.dat mum.ins
‘It is hard for the child to be dressed by the mother.’ (intended)
Secondly, by having a non-gradient definition, the analyst is encouraged to conduct more focussed analysis both inter- and intra-linguistically. This more focussed analysis can encourage research into the reasons behind the differing behaviours of nominals. From this, language-specific definitions and descriptions can be delineated.
Barddal & Eythorsson’s analysis has not covered all the languages of the world and so we cannot say with full certainty that the first argument of a verbal predicate can be analysed as the subject in every single language – but there is certainly a significant trend. What we can say with certainty, however, is that it can be applied to the Russian language. Therefore, we will work with the definition of a subject being the first argument (NP1) of a verbal predicate in neutral, unmarked constructions. Having assumed the definition of a subject in terms of argument structure, we will use this to focus our analysis on Russian dative subjects and the reasons behind their properties and behaviours. We shall refer to all such nominals found in NP1 and marked with the dative case as dative subjects.
There has thus far been no universal agreement as to the status of dative subjects in Russian. Most discussion on the topic has, however, taken an unviable nominative-centric stance. Sigurdsson (2002), for instance, believes that there are no dative subjects in Russian – in contrast to Icelandic dative subjects which he believes to behave syntactically like nominative subjects. However, behaving like a nominative nominal should not be considered a prerequisite for subjecthood since not all subjects appear in the nominative case. In contrast, Moore & Perlmutter (2000) consider Russian to be a mixed language containing consistent indirect objects, I-nominals (i.e. logical dative subjects that are in fact surface indirect objects) and true dative subjects.
Data shows there are two different types of dative subject-like nominals. Referring specifically to Russian, Moore & Perlmutter (2000) account for this difference by differentiating between what they refer to on the one hand as true dative subjects, and on the other as I-nominals, exemplified in (6) and (7) respectively. They distinguish between two types of I-nominal: governed (found with predicates such as надо, нужно, жаль, нравиться) and productive(found in semi-passive constructions).
(6) Borisu ne rabotat’ u sebja doma.
boris.dat neg work.inf at self at-home
‘It’s not (in the cards) for Boris to work at his own place.’
(Moore & Perlmutter, 2000: 378)
(7) Borisu ne rabotaet-sja u sebja doma.
boris.dat neg work.3sg-refl at self at-home
‘Boris can’t seem to work at his own place.’
(Moore & Perlmutter, 2000: 378)
According to Moore & Perlmutter (2000), the differences between dative subjects and I-nominals lie in their divergent syntactic properties. The main three distinctions they address are as follows:
- I-nominals do not determine subject-predicate agreement, whereas nominative subjects do. While the verb agrees with the subject in (8), in the I-nominal construction in (9), the verb is in the neutral 3SG form.
(8) Boris xorošo spit.
boris.nom well sleep.3sg
‘Boris sleeps well.’
(9) Borisu xorošo spit-sja.
boris.dat well sleep.3sg-refl
‘Boris sleeps well.’
- Unlike dative subjects of infinitives, I-nominals cannot occur in control contexts, such as in purpose clauses with чтобы. The dative subject in (10) is here covert PRO, as evidenced by the dative agreement on ‘odnomu’, since the case of such semi-predicatives agrees with that of their antecedents. An attempt to produce a purpose clause where an I-nominal (in this case, one in a semi-passive construction, also referred to by Moore & Perlmutter, as a productive I-construction) participates in such a control construction produces an ungrammatical utterance, as illustrated in (11).
(10) Borisi sdelal vse vozmožnoe, [čtoby PROi
boris.nom did all possible in-order PROi
‘Boris did everything possible to work alone.’
(Moore & Perlmutter, 2000: 397)
(11) *Borisi sdelal vse vozmožnoe, [čtoby PROi
boris.nom did all possible in-order PROi
ponravit’-sja èti ljudi.]
like.inf-refl these people
‘Boris did everything possible to take a liking to these people.’
(Moore & Perlmutter, 2000: 398)
- I-nominals cannot undergo raising, whereas dative subjects can. One type of raising construction in Russian involves infinitival complements of начинать/начать. When the raising predicate is an infinitive, the raisee in the matrix clause – ‘im’ in (12) – will be dative with an agreeing covert dative subject in the embedded clause. This can be seen in the agreement on ‘odnim’, which must be clause-bounded. In contrast, as illustrated in (13), the semi-passive construction cannot appear as an infinitval complement of načat’.
(12) Im ne načat’ rabotat’ odnim.
they.dat neg begin.inf work.inf alone.dat
‘It’s not (in the cards) for them to begin to work alone.’
(Moore & Perlmutter, 2000: 399)
(13) *Im ne načat’ rabotat’sja lučše.
they.dat neg begin.inf work.3sg-refl better
‘It’s not for them to begin to work better.’
(Moore & Perlmutter, 2000: 399)
Moore and Perlmutter claim that the differences between dative subjects and I-nominals can be attributed to the fact that while a dative subject can be analysed as the surface subject in surface structure, an I-nominal is the surface indirect object. They further assert that to be a dative subject in Russian a nominal cannot just be dative-marked and display some subject behaviours – it must be proven that it is not an I-nominal. Sigurdsson (2002: 693) notes that dative I-nominals are subject-like in the sense that they bear a thematic role that is normally encoded as a subject. Roles typically associated with subjects cross-linguistically include Patient, Theme and G/S/L (Goal, Source, Location). Other linguists include Instrument and Beneficiary – among others such as Experiencer.
In contrast to Moore and Perlmutter, Grillborzer (2011) differentiates putative dative subject nominals according to their semantics: while in the first group the dative nominal is an experiencer of state expressed by the predicate word (in utterances such as ‘мне холодно’, ‘I am cold’ or ‘мнетрудно‘, ‘it is hard for me’), the other involves the ‘interaction of the modal meaning with the meaning of the infinitive’. Upon analysis of the constructions which Grillborzer includes in the former group (namely the dative experiencer, dative modal and semi-passive constructions), it becomes clear that they are only found with finite verbs conjugated in 3SG (namely either быть, ‘to be’, or verbs with the reflexive suffix –ся, such as нравиться, ‘to like’). Conversely, the free infinitive construction and the constructions of potential and constructions of anteriority are found only in non-finite contexts with infinitives.
The function of dative subjects seems to be entwined in the semantics of the dative case and the way that it changes its behaviour according to the finiteness of its grammatical context. Fortuin (2006: 336) offers a Russian-specific description of the dative case:
‘The meaning of the Russian dative can be analysed in terms of notions like ‘recipient’, ‘experiencer’, etc. Typical for many uses of the dative is that there is some force directed at the participant expressed by the dative noun, and that this participant is (potentially) affected by this force, often resulting in a dynamic scene (‘receiving’, ‘coming into effect’, ‘experiencing’).’
More concisely, it seems that the dative case generally marks a goal. The dative nominal is typically the receiver or experiencer of either a literal object or abstract concept, as illustrated in (14) and (15) respectively. In (14), Tamara is the goal of her friend’s physical letter, in (15) she is the target of the abstract concept of difficulty.
(14) Podruga napisala pis’mo Tamare.
friend.nom wrote letter.acc tamara.dat
‘A friend wrote Tamara a letter.’
(15) Tamare trudno bylo napisat’ otvet podruge.
tamara.dat hard.adv was write.inf reply.acc friend.dat
‘It was hard for Tamara to write a reply to her friend.’
Interestingly, the dative can change its function according to its grammatical context while still retaining this underlying concept of a ‘goal’. Kuryłowicz (1964) differentiates between grammatical cases and concrete cases and is inconclusive on how to categorise the dative, regarding it as having both semantic and syntactic function. While grammatical cases have a primarily syntactic function (such as the accusative case, which denotes objects), concrete cases have a primarily semantic function (such as the instrumental case, which can add adverbial semantic information). Kuryłowicz describes the dative as a concrete case, yet notes that it takes on a structural function in non-finite contexts. This can be observed in Russian: In (16), the dative functions as a grammatical case, denoting the structural concept of an indirect object (he is receiving the present). Whereas in (17), the dative’s function is more semantic in that it denotes a modal meaning (he must find a present). In both utterances we can still observe the idea of the dative nominal being the target of something, in this case a gift or the concept of modality.
(16) Ja dal emu podarok.
i.nom gave him.dat present.acc
‘I gave him a present.’
(17) Emu najti podarok.
he.dat find.inf present.acc
‘He has to find a present.’
Chomsky (1981) similarly makes a distinction between ‘structural’ and ‘inherent’ case. While structural case ‘is a structural property of a formal configuration’, inherent case ‘is presumably linked to θ-role’ (Barddal, 2011: 621). This suggests the function of the dative in non-finite contexts is semantically linked to theta-roles. Furthermore, that in finite contexts the dative retains its function of marking grammatical information (i.e. indirect objects). The link between the dative and theta roles will be discussed further in Section 4, but for now it suffices to note the link between the two.
As mentioned above, the constructions appearing in Russian with dative-marked nominals can be split into two groups – one where the nominals are the first argument of the verbal predicate, and one where this is not the case. Within this latter group, there is strong evidence to suggest that this NP1 slot is filled not by the dative nominal, but instead by little pro. In light of this, such constructions cannot be classified as dative subject constructions, since the subject is in fact pro. This applies to the following constructions:
- Semi passive constructions (drefl) consist of a dative nominal + qualitative adverb/negation + 3SG reflexive intransitive verb. Moore & Perlmutter (2000) refer to them as the ‘productive I-construction’. The construction operates within a limited inventory of lexemes – unaccusative predicates, for example, do not participate. The negative form is the most widespread, but a positive form is also possible, exemplified in (18) and (19) respectively. However, the positive form is rare and in most cases implies a qualitative adverb such as хорошо, ледко, плохо, etc.
(18) Ljudi tam xorošie, prosto im
people.nom there good.nom simply them.DAT
‘People there are good people, they just don’t seem to be able to live without having to face challenges.’
(19) Zdec’ emu na redkost’ xorošo rabotalos’.
here him.dat surprisingly well work.3sg.past
‘He could work surprisingly well here.’
- Dative experiencer constructions (dpi) comprise an optional dative nominal + predicative of state + infinitive, as shown in (20). This construction is found with a limited inventory of lexemes including verbs such as нравиться, short adjectives such as нужен and experiential predicates such as трудно. Adverbial and adjectival predicates are tensed with byt’.
(20) Ø Trudno Ø predstavit’ sebe čto-libo bolee
Ø.dat hard be.Ø imagine.inf self anything more
‘It’s hard to imagine anything more deplorable.’
Testelec (2001) describes the dative-marked participant as ‘“получателем информации” или “носителем непроизвольного чувства”’, ‘the receiver of information or the carrier of involuntary feelings’ (Grillborzer, 2014: 73).
- Dative modal constructions (dmi) involve what Hansen (2001) terms modal content words, such as надо, нельзя and можно. More specifically, they are constructed as follows: dative nominal + modal auxiliary word + tensed byt’ + infinitive + arguments, as can be seen in (21).
(21) Tebe zdes’ nel’zja Ø kurit’.
you.dat here forbidden be.Ø smoke.inf
‘You may not smoke here.’
All of these constructions appear only with 3SG verbs. Semi-passive constructions are conjugated only in 3SG and when быть is overt, it too is found only in 3SG neuter form. This fact greatly suggests the presence of little pro in the subject position rather than the dative nominal, since – as we can see from utterances where dative subjects occur with infinitives –dative subjects do not always trigger verb agreement. Soschen (2003) discusses convincing evidence for pro being overtly expressed in Russian as the nominative neuter 3SG pronoun ‘oно’, ‘it’. Indeed, an example Soschen cites from colloquial Russian, illustrated in (22a), is a dative experiencer construction. If we put this utterance into the past tense to make быть overt, as expected, we see still see neuter 3SG agreement on ‘bylo’ in (22b). (22) could also be reformulated into (23) where pro is not realised as ‘ono’, but instead realised as zero.
(22a) Ono (*emu) Ø i duraku ponjatno.
it.nom (it.dat) be.Ø and idiot.dat understandable.adv
‘It goes without saying.’
(22b) Ono bylo i duraku ponjatno.
it.nom was.3sg.neut and idiot.dat clear.adv
‘It went without saying.’
(23) Ø Duraku bylo ponjatno, čto
pro.Ø idiot.dat was.3sg.neut clear.adv what
‘The idiot knew what he had to do.’
Furthermore, we can observe a difference in meaning in experiential predicates when pro is and is not the subject of an utterance. As Soschen (2003: 52) argues, postulation of pro accounts for this difference in meaning. Comparing (24) and (25), the English translations of both are ambiguous, but in Russian with ‘pro’ in the nominative in (24), Tamara is feeling the cold from an external source, whereas with ‘Tamara’ in the nominative in (25), it is her skin that is cold. With pro assuming the subject function, the dative nominal is left to function as a thematic Experiencer of the semantic content of the adverb. Since the NP1 slot has been filled, the dative nominal cannot be analysed as a subject.
(24) pro Tamare Ø xolodno.
pro.nom. tamara.dat be.Ø cold.adv
‘Tamara feels cold.’
(25) Tamara Ø xolodna.
tamara.nom be.Ø cold.adj
‘Tamara is cold.’
The dative modal and semi-passive constructions are similarly constructed in that we also see a 3SG verb – as demonstrated in (26) and (27) – allowing us to again posit the presence of pro.
(26) pro Tamare ne rabotaet-sja.
pro.Ø tamara.dat neg work.3sg-refl
‘Tamara can’t seem to work.’
(27) pro Tamare nado Ø rabotat’.
pro.Ø tamara.dat must be.Ø work.inf
‘Tamara must work.’
In summary, the dative nominals in constructions belonging to this group cannot be subjects because of the postulated presence of pro and so must be removed from the dative subject canon.
The second group of constructions, whose dative nominals are found in NP1, are all infinitival constructions and are consistent with hypotheses that state that the dative receives a more semantic function in non-finite contexts. This semantic function in the case of Russian dative-infinitival constructions is the denoting of modality. The following constructions make up the second group and can be considered dative subject constructions. The free infinitive has been retained from Old Russian, but Modern Russian also includes some innovations, namely the constructions of potential and construction of anteriority.
The dbi (also referred to across the literature as DIN/DIM/DMI) is made up of a dative nominal + tensed byt’ + infinitive + arguments, as illustrated in (28). The DS is an argument of infinitive. This is a productive construction that expresses deontic modality.
(28) Tamare Ø žit’ v obščežitii.
tamara.dat be.Ø live.inf in dormitory
‘Tamara has to live in the dormitory.’
Constructions of potentialhave been very comprehensively handled by Rappaport (1986) and are so named due to their construction of byt’ + K-word + infinitive. (The K-word being the equivalent of an English Wh-word, i.e. a question word such as кто, когда, etc.) They exist in a negative and affirmative form (exemplified in (29) and (30) respectively), though the negative form has received more attention due to its controversial structure. Both the negative and affirmative forms express the possibility or potential to do something – and both shall therefore be referred to together as constructions of potential. The positive form, however, is no longer productive in Modern Russian.
(29) Mne negde spat’.
me.dat nowhere sleep.inf
‘I have nowhere to sleep.’ (lit. to me there is nowhere to sleep)
(30) Mne est’ gde spat’.
me.dat is where sleep.inf
‘I have somewhere to sleep’ (lit. to me there is somewhere to sleep)
The structure of this construction is greatly debated (see Rappaport, 1986 for discussion) and it has been argued that the dative subject is the argument of different elements of the construction.
Grillborzer refers to the positive form as the ‘экзистенциальная конструкция’ (‘experiential construction’), and it seems to function like a possessive dative. On a similar note, Rappaport (1986: 2) notes the concurrent structural parallelism between the prepositional possessive ‘у меня есть’ (‘I have’) and constructions of potential.
constructions of anteriority (COAs) are constructed as in (31), with the order of the clauses not being fixed. This can be applied to an example such as (32), where ‘idti’ functions as Verb1 and ‘žili’ as Verb2.
(31) [conjunction of anteriority + Verb1] [Verb2]
(32) [Pered tem, kak mne idti1 v školu,] [my
before me.dat go.inf to school we.nom
s babuškoj žili2 u Orlovoj.]
with grandma.ins lived at Orlova.gen
‘Before I started school, my grandma and I lived with Ms. Orlova.’
(Fortuin, 2006: 340)
Such constructions have the meaning ‘a period in time prior to the occurrence of situation 1, occurrence of situation 2 is the case’. For example, in (32) ‘in a period of time before I started school, I lived with Ms. Orlova’. The dative subject is here an argument of the infinitive.
Though these constructions also occur with finite predicates, dative subjects are found only with non-finite predicates. Three conjunctions occur with such non-finite predicates: перед тем, как; до того, как and прежде чем, как. The dative subject, while highly infrequent in all constructions of anteriority, is found most infrequently with предже тем, как and even then, is not accepted by all speakers.
Upon comparison of the semantic functions of nominative and dative subjects, it appears that the concept of thematic roles may reveal motivations behind case selection – especially in non-finite constructions containing a dative subject. The selection of dative case marking for dative subjects is plausibly linked to the dative’s associations with the ‘goal’ of semantic meaning, which are also inherent to the thematic role of ‘Experiencer’. All dative subjects in Russian can be described as thematic Experiencers of modality.
It is widely accepted that a verbal predicate has a set number of arguments to satisfy and satisfies these according to a thematic hierarchy. It is generally agreed that the Agent role should be the highest ranking in the hierarchy, but there is no consensus as to the subsequent ordering of thematic roles.
Cross-linguistically, the notions of grammatical subject (traditionally thought of as nominative) and thematic Agent are often confused – yet they are distinct. The term ‘Agent’ implies a sense of causation or action in its enactment of the verb, but this is not necessarily the case even for a nominative subject. This is especially the case in utterances where the nominative nominal is in fact a Patient/Theme, such as (33) and (34):
(33) The office was burgled yesterday.
(34) Kniga byla pročitana.
book.nom was read.pp
‘The book was read.’
A subject in Russian can be described as the first argument (NP1) of the verbal predicate in the surface structure that is assigned a thematic role which speakers associate with a subject according to the following hierarchy: Agent > Experiencer > Patient/Theme. This means that in the absence of an Agent, an Experiencer is selected as the first argument of the verb and in the absence of both an Experiencer and an Agent, a Patient/Theme is selected. This can be formulated as:
(35) Subject = [NP]1 = Agent > Experiencer > Patient/Theme.
Unaccusative constructions, for example, lack both an Agent and an Experiencer. This results in a nominative Theme assuming the subject position, as in (36).
(36) Pojavilsja mal’čik.
‘There appeared a boy.’
When comparing the semantic context of nominative and dative subjects in Russian, a pattern emerges regarding their thematic roles. Nominative subjects can consistently be analysed as Agents, Patients, or Themes, whereas dative subjects can consistently be analysed as Experiencers. The grammatical case of a subject therefore denotes its thematic role as identified by speakers.
It appears that the interaction of dative nominal and infinitive generates a modal construction. If we examine the semantics of all of the dative-infinitival constructions in this group, they all have a sense of modality. A free infinitive construction such as (37) has a clear deontic modal meaning. Constructions of potential also fairly clearly express the (im)possibility of an action, as in (38) where the speaker has no possibility to sleep anywhere.
(37) Mne spat’.
‘I have to sleep.’
(38) Mne negde spat’.
me.dat nowhere sleep.inf
‘I have nowhere to sleep.’
At first glance, the case of the modal semantics of constructions of anteriority is slightly less clear cut. But Fortuin (2006) offers convincing arguments that such constructions with a dative subject express an inevitable happening:
‘The [construction of anteriority with a dative subject] expresses that the infinitive situation is due to be realised, for example because there is a plan according to which the situation has to occur, or because of the inevitable way things go.’ (Fortuin, 2006: 354)
If we examine (39) and (40), they both depict a situation inevitable in our world – the rising of the moon and the turn of the seasons. It is a situation that must be realised because there is no other possible outcome.
(39) pered tem, kak vzojti lune […]
before rise moon.dat
‘before the moon rises […]’
(40) Osen’ doživala poslednie denjočki, pered tem kak
autumn.nom lived out last days before
‘The last days of autumn played out before winter fell.’
Though, such constructions are not limited to utterances relating to the natural world. (41) indicates a time before the speaker had to start school. Starting school is something every child is generally expected to do and therefore there is logically no other possible outcome – it is an action that must and will occur.
(41) Pered tem, kak mne idti v školu, my
before me.dat go.inf to school we.nom
s babuškoj žili u Orlovoj.
with grandma.ins lived at Orlova.gen
‘Before I started school, my grandma and I lived with Ms. Orlova.’
(Fortuin, 2006: 340)
Indeed, a dative subject cannot be inserted into constructions that do not have a modal meaning. Fortuin (2006: 337) cites the example of the proverb ‘много знать, мало спать’, ‘to know a lot, you must sleep a little’. A dative nominal could only here be inserted into the second clause, due to the proverb in full being conditional in nature: ‘Кто хочет много знать, тому надо мало спать’. Further to this, a dative NP can be inserted into other constructions that do have a modal meaning without altering the semantics. When a dative subject is inserted into the free infinitive construction in (42), the utterance still expresses the fact that smoking is not allowed.
(42a) Zdes’ ø ne kurit’.
here NP.dat.ø neg smoke.inf
‘One may not smoke here.’
(42b) Zdes’ mne ne kurit’.
here me.dat neg smoke.inf
‘I may not smoke here.’
In some contexts, dative subjects do not always have to be realised in these dative-infinitival constructions. When realised, a dative subject can be utilised to express modality on a specific subject. Realised as zero, the utterance has a general meaning. Indeed, if we compare (41-43) with and without a dative subject ((a) and (b) respectively), this becomes clear:
(43a) Čto ø poest’ na zavtrak, kogda nečego
what NP.dat.ø eat.inf for breakfast when nothing.gen
net v xolodil’nike?
neg in fridge.prep
‘What should you eat for breakfast when there is nothing in the fridge?’
(43b) Čto mne poest’ na zavtrak, kogda nečego
what i.dat eat.inf for breakfast when nothing.gen
net v xolodil’nike?
neg in fridge.prep
‘What should I eat for breakfast when there is nothing in the fridge?’
A dative subject can thus be described as a facultative Experiencer of modality, realised as zero with a general meaning and surfacing when the speaker wishes to express a specific Experiencer.
Having established that dative subjects are linked with the concept of modality, a logical question that follows this is: Why does this association exist? In the case of the dative subject constructions, this seems to lie in the modal semantics of the infinitive, which it has acquired historically though speaker reanalysis of its syntactic role in purpose clauses and its associations with the cross-linguistic phenomenon of modal ‘have to’ constructions.
The development of the modal semantics of the infinitive appears to have its roots in the development of possessive constructions into purpose clauses. Old Russian had two ways of expressing possession: the ‘possessive dative’ and the ‘prepositional possessive’, illustrated in (44) and (45) respectively. As Jung (2010) notes, both are similarly constructed with an oblique NP + ‘byt’’ + nominative theme (NPOBLIQUE – Ø – NPNOM). The dative possessive eventually becomes non-productive (though vestiges of it remain in phrases such as ‘мне 22 года’, ‘I am 22’). Ickovich (1982) distinguishes their function thus: while the bare dative construction “points to the designation of the object (property, characteristic), […] the prepositional phrase denotes possession”. We can see this exemplified below: Both utterances express the idea of having a husband, but while (44) refers more to the status of being a husband, in contrast, (45) expresses physical possession of a husband. The dative indicates more of an abstract relation than actual possession.
(44) On ø mne muž.
he be.ø me.dat husband.nom
‘He is my husband.’
(45) U menja ø muž.
I have (lit. ‘by me.acc’) be.ø husband.nom
‘I have a husband.’
In addition to the prepositional possessive, the possessive dative also has structural links with modal constructions expressing necessity. In English, as well as a range of other languages, we find the ‘have to’ construction expressing necessity in utterances such as: ‘We have to go!’. Jung (2010) draws structural comparisons between the possessive dative and such modal ‘have to’ constructions in Russian (i.e. the free infinitive construction) in their parallel structure of oblique NP + byt’ and it seemsthat their structural development is linked. The development of possessive constructions into those expressing necessity or possibility is a cross-linguistic phenomenon and best illustrated by what Heine (1993) calls the purpose schema, exemplified thus:
(46) 1. I have a letter (existential/possessive construction)
2. I have a letter to write (purpose construction, e.g. infinitive of purpose)
3. I have to write a letter (used with transitive verbs)
4. I have to write (used with intransitive verbs)
Using Modern Russian translations of the utterances in (46) for the sake of example, Stage 1 in Russian would have involved a prepositional phrase along the lines of (47a), progressing to stage 2, exemplified in (47b) and eventually progressing through stage 3 to stage 4, exemplified in (47c) and (47d) respectively.
(47) a. U menja pis’mo.
I have letter.nom
b. U menja pis’mo pisat’.
I have letter.nom write.inf
c. Mne est’ pis’mo pisat’.
me.dat is letter.nom write.inf
d. Mne pisat’ pis’mo.
me.dat write.inf letter.nom
At stage 2, there is evidence to suggest a reanalysis of the infinitive + object to denote modality in purpose clauses. In Old Russian, necessity was expressed with an infinitive of purpose, as illustrated in (48).
(48) takova pravda uzjati Rusinu
such rights.nom take.inf Russian.dat
‘these are the rights a Russian must enjoy’
Holvoet (2003) discusses convincing evidence, supported by data from other Baltic languages, that the nominative subject in this construction (‘takova Pravda’) was reanalysed as an object in the embedded clause. The infinitive was therefore analysed as requiring an object – making it plausible that the combination of infinitive + object was reanalysed as denoting necessity (a type of modality).
The emergence of constructions of potential seems to be linked to stage 3, which involves the selection of a construction with a dative nominal over one with a prepositional phrase. Constructions of potential (identical in structure to constructions at stage 3) were not present in the Old Russian inventory of dative subject constructions. They seem to be an innovation born of stage 3 of the purpose schema, where the possessive and the modal appear to blend: Jung (2010) goes on to argue that the structural similarity between the possessive constructions and free infinitive constructions serves to help blur the semantics between the possessive and the modal. Rappaport (1986) also notes the structural parallelism between constructions of potential and ‘у меня есть’ possessive constructions. The decline in the use of dative possessive constructions would leave the dative in this context open to exaptation. This fact, along with the semantic blurring between the possessive and the modal, would underpin stage 3 – where, after a period of ambiguity, the dative nominal is selected over the prepositional phrase. This is likely supported by the dative’s function of expressing abstract links between nominals, while the prepositional phrase rather tends to denote physical possession.
This combination, then, of dative + infinitive has historical evidence to support its development to denote modality. The appearance of a full verb would not have the same links with modality that an infinitive does, meaning dative subjects in such constructions do not appear with finite verbs. German also utilises a bare infinitive + ‘to be’ to express obligation or necessity, as illustrated in (49). This cross-linguistic link between the infinitive and the modal across language families also suggests some kind of Indo-European inheritance.
(49) Das Formular ist auszufüllen.
the.nom form be.3sg fill out.inf
‘The form should be filled out.’
We have seen that previous definitions of subjecthood have been inadequate in their focus on the so-called ‘subject properties’ and in so doing have hindered a subjective and principled analysis of putative subject nominals and their behaviours. In regarding the first argument (NP1) of a predicate in unmarked sentences as the subject of an utterance, we are forced to view subjecthood not as a gradient concept. In light of this, the canon of dative subject constructions must be reconsidered, since the dative modal, dative experiencer and semi-passive constructions cannot be considered to exhibit true dative subjects since, due to the presence of little pro, the dative-marked nominals in such constructions cannot be considered NP1. This leaves only the free infinitive construction, constructions of anteriority and potential containing dative-marked nominals as Russian dative subject constructions.
When observing and analysing the behaviours of the dative subject nominals in these constructions, we can describe them as thematic experiencers of modality. Our analysis also strongly suggests that the subject status of Russian dative-marked nominals seems to be dependent on whether or not they are found in non-finite contexts. When found in non-finite contexts, the dative nominal (functioning as a thematic experiencer) and infinitive work together to create the modal semantics of such utterances. Such constructions plausibly stem from the possessive dative, which itself fell out of use, but developed into the modern canon of dative subject constructions. This has implications for the debate surrounding the argument structure of constructions of potential since it suggests that the dative nominal was first analysed as the first argument of the infinitive.
Having started to analyse Russian dative subjects, this non-squishy definition in terms of argument structure can now start to be applied to other putative subject nominals in both Russian and other languages in an attempt to dig deeper into the reasons behind subject nominal’s behaviours – which can often be found in their historical development.
Bailyn, John F. 2011. The Syntax of Russian (Cambridge Syntax Guides). Cambridge: CUP.
Barddal, Johanna. 2011. Lexical vs. structural case: a false dichotomy. Morphology 21. 619-654.
Barddal, Johanna & Thorhallur Eythorsson. To appear. What is a Subject? The Nature and Validity of Subject Tests. In J. Barddal, S.M. Carey, T. Eythórsson & N. Pat-El (eds.), Non-Canonically Case-Marked Subjects within and across Languages and Language Families: The Reykjavík–Eyjafjallajökull Papers.
Besters–Dilger, Juliane, Ana Drobnjaković & Björn Hansen. 2009. Modals in the Slavonic languages. In Hansen, Björn & Ferdinand de Haan (eds.), Modals in the Languages of Europe: A Reference Work. Berlin: De Gruyter. 167–199.
Bresnan, Joan. 2001. Lexical-functional syntax. Oxford: Blackwell.
Brown, Sue. 1999. The Syntax of Negation in Russian: A Minimalist Approach. Stanford: CSLI Publications.
Eythorsson, Thorhallur & Johanna Barddal. 2005. Oblique Subjects: A Common Germanic Inheritance. Language 81.4. 824-81.
Fleischer, Nicholas. 2006. Russian Dative Subjects, Case, and Control. Ms., UC Berkeley.
Fortuin, Egbert. 2006. On the Use of Dative Subjects in the Construction of Anteriority in Russian. Russian Linguistics 30.3. 321-357.
Grillborzer, Christine. 2014. Glava IV: Status pervogo datel’nogo aktanta v sovremennom russkom jazyke. In Sintaksis konstrukcij s pervym datel’nym aktantom – Sinchronnyj i diachronnyj analiz. Regensburg: University of Regensburg (PhD Thesis). 51-123.
Grillborzer, Christine. 2011. Dative subjects in Russian: The evolution of subject-like properties from Old to Modern Russian. http://org.uib.no/iecastp/IECASTP/Grillborzer_Bergen.pdf [accessed 4 July 2017].
Gutiérrez, Juan Pablo Mora. 2001. Thematic Hierarchies. Estudios de lingüística del español 11. http://elies.rediris.es/elies11/cap51221.htm [accessed 4 July 2017]
Holvoet, Axel. 2003. Modal constructions with ‘be’ and the infinitive in Slavonic and Baltic. Zeitschrift für Slawistik 48.4. 465-480.
Johnson, David E. 1977. On Keenan’s Definition of Subject Of’. Linguistic Inquiry 8.4. 673-692.
Jung, Hakyung. 2010. Innovations in the Rise of the Dative-Infinitive Modal Sentence in Russian. Zeitschrift für Slawistik, 55-4. 377-399.
Jung, Hakyung. 2009. Null Prepositional Complementizers and the Dative of Obligation in Russian. Proceedings of Formal Approaches to Slavic Linguistics 17. Ann Arbor: Michigan Slavic Publications. 64-80.
Keenan, Edward L. 1976. Towards a Universal Definition of ‘Subject’’. In Charles N. Li (ed.) Subject and Topic. New York: Academic Press. 303-333.
Levin, Beth. 2006. Lexical Semantics and Argument Realization V: Thematic Hierarchies and Semantic Prominence. http://web.stanford.edu/~bclevin/dgfs06thier.pdf [accessed 4 July 2017].
Madariaga, Nerea. 2011. Infinitive clauses and dative subjects in Russian. Russian Linguistics 35.3. 301-329.
McCloskey, Jim. 1997. Subjecthood and Subject Positions. In Liliane Haegeman (ed.), Handbook in Generative Syntax. Dordrecht: Kluwer Academic. 197-235.
Moore, John & David M. Perlmutter. 2000. What Does It Take to be a Dative Subject?. Natural Language & Linguistic Theory 18.2. 373-416.
Rappaport, Gilbert C. 1986. On a Persistent Problem of Russian Syntax: Sentences of the Type ‘Mne Negde Spat’. Russian Linguistics 10.1. 1-31.
Sigurdsson, Halldór Ármann. 2002. To Be an Oblique Subject: Russian vs. Icelandic. Natural Language & Linguistic Theory 20.4. 691-724.
Soschen, Alona. 2003. On Subjects and Predicates in Russian: Syntax/Semantics Interface. Florida: Dissertation.com.
Sridhar, S.N. 1979. Dative subjects and the notion of subject. Lingua 49. 99-125.
Vincent, Nigel & Thórhallur Eythórsson. The rise and fall of dative subjects. http://www.let.leidenuniv.nl/ulcl/events/compdiachr/VincentEY.pdf [accessed 4 July 2017].
van Kampen, Jacqueline. 2006. Subjects and the (Extended) Projection Principle. In Keith Brown (ed.), Encyclopedia of Language and Linguistics, 2nd ed. Vol. 12. Oxford: Elsevier Science. 242-248.
Word count: 7361 words
 Definitions of the constructions in these two groups mentioned here can be found in Section 3.
 Bresnan (2001: 304) defines argument structure as ‘an interface between the semantics and syntax of predicators (which we may take to be verbs in the general case) …’. Argument structure causes a lexical head to generate argument positions in syntactic structure. It encodes information such as the number of arguments a lexical item takes and their syntactic expression. We will borrow Bailyn’s (2011: 4) notation of argument structure, where he describes argument structure as follows: [NP1 ___ NP2, NP3, (PP4)], with NP1 representing the first argument of the verbal predicate, NP2 the second, and so on.
 The English translations of (6-7) do not necessarily show the fine distinction in meaning between each pair. While the dative subject construction in (6) implies it is external circumstances not allowing Boris to work or sleep, in the semi-passive construction in (7) it is an internal feeling that is preventing him from doing so.
 This is a cross-linguistic property of the dative case: Jakobson (1953), Dative signifies ‘goal’ or ‘marginality; Janda (1993), in Czech, dative signifies receiver, experiencer or competitor.
 Little pro (as opposed to big PRO, which can never be realised) is a null pronoun-like NP without phonetic properties which replaces an overt pronoun and may or may not be realised. It can be found in the subject position and can be assigned case.
 For a more in-depth discussion of the characteristics of each construction, consult Grillborzer, 2014: 69-75. Grillborzer’s work (2014; 2011) also provides further discussion on the characteristics, behaviours and development of dative nominals from Old Russian to the present day.
 A large amount of literature exists on the topic of the ordering of thematic hierarchies. For an example overview of proposals that have already been made, see the list collected by Levin and Rappaport (1996) in Gutiérrez (2001).
 A Patient/Theme can become a subject, for example, when an utterance is passivised and the Agent (the original subject) is made to function as the Instrument, as in (i), where ‘dom’ raises to subject and ‘Tamara’ is made to function as the instrument. Functioning as an instrument, an Agent will receive instrumental-marking, but when functioning as a subject, it will receive nominative-marking. Whereas, as an object, a Patient/Theme will receive accusative-marking and as a subject will receive nominative-marking.
(ib) Tamara stroila dom.
tamara.nom-agent built house.acc-patient
‘Tamara built the house.’
(ib) Dom byl’ postroen Tamaroj.
house.nom-patient was built.pp tamara.ins-instrument
‘The house was built by Tamara.’
Not only accusative Patients/Themes can be raised to the first argument position: Fowler (1996) has identified the phenomenon of ‘oblique passivisation’, whereby Theme arguments in oblique cases are nominative-marked when passivised. This is exemplified in (ii), where the genitive nominal ‘celi’ is raised to subject in (iib).
(iia) Tamara dostigla celi.
tamara.nom-agent achieved goal.gen-patient
‘Tamara achieved her goal.’
(iib) Cel’ byla dostignuta Tamaroj.
goal.nom-patient was achieved tamara.ins-instrument
‘The goal was achieved by Tamara.’
 It should be noted here that there is some speaker variation and not all speakers accept the insertion of a dative nominal into such constructions.
 See Fortuin (2006: 336-339) for a convincing argument that the generic nature of utterances where a dative NP can be inserted blocks the realisation of a dative nominal.
 See Jung (2010) for more detailed discussion of the structural development of free infinitive constructions and their roots in possessive constructions.
 The purpose schema and its relation to Russian is discussed in more detail in Holvoet, 2003: 466.
 The historical evidence discussed here suggests that the dative subject in constructions of potential was initially analysed as an argument of the infinitive, with the accusative nominal analysed as an object of the infinitive. The construction may have undergone further speaker reanalysis over time and it is beyond the scope of this dissertation to go into an in-depth discussion of the construction of constructions of potential, but it suffices to note the apparent origins of their argument structure.