You are on page 1of 29

High Datives in Polish*

Natalia Cichosz

Abstract
In this paper I will propose an analysis of dative nominals that occur in the highest sentence-initial
position in a number of structures in Polish. Contrary to proposals inspired by Pylkknens (2008)
applicative model, I will argue that high datives in Polish should be analysed as adjuncts that
select the structure that they combine with on the basis of its eventive semantics. In order to
account for syntactic and semantic properties of the dative DP, I will propose that the dative
should be interpreted as a participant in the event it is associated with but not in control of. This
out-of-control semantics will be expressed either when the dative is identified with the subject
argument of the event it is associated with or when it is affected by that event. The choice of the
appropriate interpretation for a given structure will be regulated by a context sensitive rule
responsible for co-indexation of the dative with the subject argument available in the structure that
it combines with. In the absence of an appropriate argument in the structure, it will be proposed
that the dative receives a default affectedness interpretation that allows it to be interpreted as a
participant in the event. The adjunct status of the dative will be confirmed by evidence coming
from binding and secondary predication data. It will be demonstrated that the analysis proposed in
this paper is able to correctly predict the semantic and syntactic properties of dative nominals and
structures in which they occur.
Keywords: dative, Polish, out-of-control, affectedness, reflexive

1 Introduction

The distinction between arguments and adjuncts has commonly been considered one of the
key issues forming part of most current formal linguistic theories. Even though it is difficult
to provide a definition of these notions, there are certain semantic and syntactic criteria that
help to determine whether a given category should be classified as an argument or an adjunct.
For instance, arguments, but not adjuncts, are expected to receive an interpretation that
belongs to a limited list of semantic functions such as agents, experiencers, goals, etc.
Syntactic criteria draw a distinction between arguments and adjuncts on the basis of, for
example, their obligatory or optional presence in the structure, movement and extraction
possibilities. As arguments are also subcategorised for by the verb and directly related to its
meaning, their omission often renders the sentence ungrammatical. Adjuncts, on the other
hand, are never obligatory, since they are not implied by the semantics of the predicate. As
they are always optional elements of the sentence, when removed they do not create any
sense of semantic incompleteness or ungrammaticality.
In the present paper I will focus on a number of Polish structures particularly
significant from the point of view of the above argument adjunct distinction. I will
demonstrate that the common assumption that a case-marked DP referring to an individual
associated with an event should be assigned the argument status can be challenged when
confronted with data coming from less widely investigated languages.

1.1 Preliminary data

* I am especially grateful to Ad Neeleman for all his advice and support. Special thanks are also due to
Klaus Abels for many valuable comments and discussions.
UCLWPL 2012 123

The structures that provide crucial evidence in support of the above claim are illustrated
below.

(1) a Jankowi dobrze si taczyo.


John.Dat. well refl danced.3sg.n.
John danced and he experienced that as pleasurable.

(2) b Jankowi si kichno.


John.Dat. refl sneezed.3sg.n.
John (accidentally) sneezed.

Jankowi zepsu si samochd Pawa.


John.Dat. broke down.3sg.m. refl car.Nom. Paul.Gen.
Pauls car broke down and John was affected by that.

(3) a Jankowi Ania zepsua samochd Pawa.


John.Dat. Ann.Nom. broke.3sg.f. car.Acc. Paul.Gen.
Ann broke Pauls car and John was affected by that.

b Jankowi Ania wyjechaa z kraju.


John.Dat. Ann.Nom. left.3sg.f. from country
Ann left the country and John was affected by that.

(4) a Jankowi te teksty atwo si tumacz.


John.Dat. these texts.Nom. easily refl translate.3pl.
To John, these texts translate easily.

b Jankowi napisay si dwie litery.


John.Dat. wrote.3pl. refl two letters.Nom.
John accidentally wrote two letters (of the alphabet).

In what follows, contrary to the assumptions of traditional argument structure theories that
would ascribe the argument status to the dative DPs in the above sentences, I will
demonstrate that even though they seem to share certain properties with arguments, dative
nominals in (1) (4) should in fact be analysed as adjuncts. In section 2, I will propose a
semantic specification for the dative and discuss the interpretations that the dative can receive
in various contexts as well as the mechanisms by which those interpretations can be assigned.
It will be demonstrated that the proposal put forward in this paper correctly predicts the
syntactic and semantic properties of dative-nominal sentences. Section 3 will discuss the
syntax of dative-nominal structures and provide further arguments in favour of the adjunct
status of the dative. Finally, I will refer to possible alternative analyses of non-argumental
datives and demonstrate that they cannot account for the properties of the dative in Polish.
In all the structures provided in (1) (4) above, the dative requires that the sentence it
combines with express an event it is associated with but not in control of. In some of these
structures, the dative is identified with the performer of the action and in cases in which it is
not associated with the subject argument, it is interpreted as an individual affected by a given
event. There are no particular requirements with regards to the morphology of structures that
can combine with the dative. However, as will be demonstrated further in the paper, the
UCLWPL 2012 124

availability of the actor argument interpretation of the dative depends on the presence of the
reflexive morpheme si1 in sentences that the dative precedes.
In (1a) and (1b) the dative DP co-occurs with an impersonal sentence that is
obligatorily composed of the neuter 3ps sg form of the verb and the morpheme si. I will
refer to the dative sentences exemplified in (1) as the dative impersonal si construction. The
sentence in (1a) also contains an obligatory experiencer adverb that refers to the manner in
which the agent perceives the action that he is performing and not to the quality of the action
itself. In both (1a) and (1b) the impersonal structure that the dative attaches to, illustrated by
(5a) and (5b) below, expresses an action that is performed by an arbitrary [+human] subject.2
The adverb in (5a), similarly to its interpretation in (1a), normally refers to the manner in
which the arbitrary subject perceives the action that he is performing.3

(5) a Dobrze si taczyo.


well refl danced.3sg.n.
One danced and they experienced their dancing as good.

b Kichno si.
sneezed.3sg.n. refl
One (accidentally) sneezed.

This interpretation of the adverb is not possible in nominative-subject sentences. In those


cases, the adverb can only refer to the quality of the action:

(6) Janek dobrze taczy.


John.Nom. well danced.3sg.m.
Not: John danced and he experienced his dancing as good.

As indicated earlier, the difference between the two dative impersonal si sentences in (1) lies
in the obligatory presence of an experiencer adverb in one of them. In both cases, however,
the dative is not in control of the event in which it participates. The nominative-subject

1
Reflexive is the term traditionally used to refer to the morpheme si. I will use this term throughout the
paper even though si has many usages some of which are not related to the notion of reflexivity.
2
Depending on the referent(s) available in the context in which the sentence is uttered, the covert arbitrary
subject of the impersonal si construction may be specified for any person, number or gender (the following
example has been adopted from Kibort (2004: 288)):

(i) Pracowao si jako nauczyciel/ nauczycielka/ nauczyciele/ nauczycielki.


worked.3sg.n. refl as teacher.m. teacher.f. teachers.vir. teachers.non.vir.
One worked as a teacher./People worked as teachers.

As will be illustrated further in the paper, the presence of such an arbitrary subject in the structure will be
necessary for the dative in sentences such as (1a) and (1b) to receive an (involuntary) agent/experiencer
interpretation.
3
Alternatively, in certain cases an experiencer adverb can express the speakers attitude towards the event
in which he is not interpreted as (one of) the actor(s):

(i) W tej firmie przyjemnie zaczyna si prac o 10.00.


in this company pleasurably start.3sg.n. refl work at 10.00
1. In this company one starts work at 10.00 and I consider this pleasurable.
2. In this company one starts work at 10.00 and they experience that as pleasurable.
UCLWPL 2012 125

version of (1), illustrated by (7) below, is underspecified with regards to whether or not the
subject is in control of the action it is compatible with a scenario according to which John
performs the action on purpose and with the one in which he does it accidentally.

(7) Janek kichn.


John.Nom. sneezed.3sg.m.
John sneezed.

For this reason, it seems that the use of the dative case in sentences such as (1b) is justified
only if the event expressed by the predicate can in theory be controlled by the actor argument.
This initial observation is confirmed by the ungrammaticality of sentences in which the
dative precedes a subject experiencer verb that describes a state that cannot be controlled4:

(8) * Jankowi lubi si Ani.


John.Dat. likes.3sg.n. refl Ann.Acc.
Intended reading: John likes Ann.

If the dative cannot be identified with the subject argument of the event described by
the sentence it combines with, it has to be interpreted as affected by that event. As illustrated
by (2) and (3) above, the affectedness dative can precede sentences that belong to various
syntactic types. In (2) the dative attaches to the anticausative structure and in (3a) (3b) it
precedes a transitive and unergative sentence, respectively.5 The obligatory affectedness
interpretation of the dative occurring in the above-mentioned sentences is due to the fact that
they all contain independent grammatical subjects with which the dative cannot be identified.
The availability of the affectedness interpretation for the dative will be discussed in the
relevant sections of this paper.

4
There is no similar restriction on the formation of impersonal si sentences that contain subject
experiencer verbs:

(i) Lubi si Ani.


likes.3sg.n. refl Ann.Acc.
One likes Ania.

Reasons for why this restriction should hold in the case of dative sentences such as (7) will be provided further
in this section.
5
An affectedness dative can combine with any type of structure illustrated in (3) which has an independent
grammatical subject. In the sentence below it precedes the impersonal no/ -to construction that contains a
covert subject which has to have a human referent. However, there are reasons to believe that there are certain
restrictions on possible subjects of impersonal no/ -to sentences that do not apply in the impersonal si
construction (but see Kibort (2004) for discussion). As a result, in the sentence below the dative cannot receive
the involuntary agent/experiencer interpretation that was available for it in sentences in which it preceded the
impersonal si construction.

(i) Jankowi zepsuto samochd Pawa.


John.Dat. broke.impers. car.Acc. Paul.Gen.
People broke Paul's car and John was affected by that.
UCLWPL 2012 126

In the middle sentence in (4a), the dative DP is normally interpreted as the performer of
the action. Similarly to the sentence in (1a), it is not in control of the manner in which it
experiences that action. However, in some cases in the middle and in the impersonal si
construction the dative can also be interpreted as affected by the action performed by an
arbitrary human subject. I will refer to some of those contexts in section 2.
The structure in (4b) is similar to (1b) in that the action happens spontaneously.
However, it turns out that for a number of reasons (4b) patterns with the middle rather than
with the impersonal si construction. For example, it does not contain a syntactically active
agent6 and due to obligatory agreement between the verb and the logical object, it can only be
used with transitive verbs.
As shown by the examples in (1) (4), the structures so far introduced in this paper are
restricted to sentences that contain high datives. These dative DPs should be distinguished
from other kinds of datives that are also widely used in Polish. The relevant examples include
datives functioning as sole or indirect objects, beneficiary and ethical datives. They are
illustrated by the sentences in (9) (11) below, respectively:

(9) Janek nie ufa swojemu lekarzowi.


John.Nom. neg trusted.3sg.m his.refl. doctor.Dat.
John didnt trust his doctor.

(10) Upiekam Jankowi ciasto.


baked.1sg. John.Dat. cake.Acc.
I baked a cake for John. or I baked a cake so that John does not have to do that.

(11) Nie bd mi znowu niemiy dla Janka!


neg be.imp. I.Dat. again unkind for John
Dont be unkind to John again + ethical dative me

Dative DPs occur also in the structure illustrated by (12) below. Even though it is not
clear at the moment how they should be analysed, it is still possible to point to some of their
properties that seem to demarcate them from dative nominals occurring in (1) (4). As

6
The following sentences illustrate the contrast between the middle and the impersonal si construction
with respect to the availability of a syntactically active subject in their structures. The impersonal si
construction can be followed by a purpose clause (i) and modified by the adverb celowo deliberately in (ii).
This is not possible in the case of middle sentences in (iii) and (iv):

(i) Taczyo si eby zadowoli publik.


danced.3sg.n. refl in order to please.Inf. audience.Acc.
One danced in order to please the audience.

(ii) Pracowao si powoli celowo.


worked.3sg.n refl slowly deliberately
One worked slowly on purpose.

(iii) *Te teksty szybko si tumacz eby zadowoli wydawc.


these texts.Nom. fast refl translate.3.pl in order to please.Inf. publisher
*These texts translate fast in order to please the publisher.

(iv) *Te buki szybko kroj si celowo.


these rolls.Nom. fast cut.3pl. refl deliberately
*These rolls cut fast deliberately.
UCLWPL 2012 127

shown below, the structure in (12) should be distinguished from those in (1) (4) on the basis
of its special morphological form. To be more precise, the predicate of (12) is formed of an
adverb7 but it is also possible to have nouns and certain non-personal verb-like elements in
this function.8 None of these featured in this role in the dative sentences in (1) (4).

(12) Jankowi byo przyjemnie/ zimno/ wygodnie/ smutno.


John.Dat. was.3sg.n. pleasurably cold.adv. comfortably sadly
John felt nice/ cold/ comfortable/ sad.

Similarly to the above structure, also beneficiary and ethical datives demonstrate properties
that differentiate them from dative DPs in (1) (4). The distinct behaviour of ethical datives
can be observed on the basis of their semantic contribution to the meaning of the structure in
which they occur and some distributional properties that are clearly different from those
demonstrated by the highest datives in (1) (4). Beneficiary datives and those that function
as sole/indirect objects are normally assumed to originate within the VP and the difference
between them and high dative DPs is related to their distribution in focus and in the
correction context. In addition, the same distinction between high and low datives can be
illustrated on the basis of their interpretation in sentences modified by the adverb ponownie
again and some reconstruction for binding data. For reason of space, I will limit my
discussion of evidence that differentiates between high and low datives to the data provided
by the again and the reconstruction for binding test.

1.2 High and low datives distinction

In sentences in which the dative receives the performer of the action interpretation, there is no
need to assume that it does not surface in its base-generated position. As it is claimed that the
dative originates so high in the structure, the constituent it combines with is expected to
occupy the position either to the left or to the right of the dative. Since in Polish the
performer of the action typically occupies the initial position in nominative-subject sentences,
when the dative is identified with the subject argument and therefore receives the performer
of the action interpretation, it will also be expected to occur sentence-initially rather than to
follow the rest of the clause. This prediction has already been confirmed by the examples in
(1a) and (1b).
However, the question of high or low attachment is particularly relevant for
affectedness datives that similarly to the lower dative in (10) are not identified with the
subject argument of the event. First of all, if beneficiary datives originate in the low position,
as opposed to high datives in (1) (4), they should only be able to occupy the sentence-initial
position in sentences such as (13) below as a result of movement:

7
There is some evidence that adverbs occurring in structures such as (12) should in fact be analysed as
non-agreeing predicative adjectives. See, for example, Kibort (2004) and references thereof.
8
This can be illustrated by the following sentences:

(i) Jankowi byo wstyd.


John.Dat. was.3sg.n. shame.m.Nom.
John felt ashamed.

(ii) Jankowi nie wolno pracowa.


John.Dat. neg be-allowed.non-personal. work.Inf.
John is not allowed to work.
UCLWPL 2012 128

(13) Marysii Janek upiek ti ciasto.


Mary.Dat. John.Nom. baked.3sg.m. cake.Acc.
John baked a cake for Mary.

The test that confirms the contrast between beneficiary and affectedness datives with
regards to the position in which they are generated in the structure involves modification of
dative-nominal sentences by the adverb ponownie again. As the adverb can take scope over
different constituents attaching in various positions, structures in which it occurs below the
dative but above the subject position can provide evidence for different attachment sites of
high and low datives. If the dative in (13) originates in the low position, in sentences such as
(14) below it should be interpreted internally to the constituent containing the adverb. This is
confirmed by the data:

(14) Jankowi ponownie Ania upieka ciasto.


John.Dat. again Ann.Nom. baked.3sg.f. cake.Acc.
Ania baked a cake for John again.

Namely, (14) is true only when the dative has already been interpreted before as the
beneficiary of the action described by this sentence. Therefore, it cannot refer to the situation
in which John would have the cake baked for him for the first time.
On the contrary, the affectedness dative in the sentence below can be interpreted
outside the scope of the adverb, which confirms that it can be generated in the highest
position:

(15) Jankowi ponownie Ania potuka wazon jego mamy.


John.Dat. again Ann.Nom. broke.3sg.f. vase.Acc. his mum.Gen.
Ania broke Johns mums vase again and John was affected by that.

Unlike the beneficiary dative in (14), the dative DP in the above sentence can be interpreted
externally to the constituent containing the adverb. For this reason, (15) can be used also
when Ania broke the vase again but John was affected by that for the first time. For example,
John did not know that it was Ania who originally broke the vase but as he was involved in
gluing the vase together, he was affected by Ania breaking it again.
The above data confirm the difference between beneficiary and affectedness datives
with respect to the position in which they are generated in the structure. The same contrast
should also be demonstrated by the reconstruction for binding data that involve structures in
which the dative DP occurring sentence-initially contains a subject-oriented reflexive
possessor swj.
As beneficiary datives are assumed to originate low in the structure, in sentences in
which they appear sentence-initially, they should be able to reconstruct and allow binding of
the reflexive possessor by the subject. This prediction is confirmed by the grammaticality of
(16):

(16) [Swojej siostrze] Janek upiek ciasto.


[his.refl. sister]Dat. John.Nom. baked.3sg.m. cake.Acc
John baked a cake for his sister.
UCLWPL 2012 129

On the contrary, high datives should not allow modification by the reflexive possessor. The
lack of a reconstruction site for the dative combining with the middle structure is confirmed
by the ungrammaticality of the following example:

(17) *Swoim autorom te teksty najlepiej si tumacz.


[their.refl. authors]Dat. these texts.Nom. best refl translate.3pl
Intended meaning: These texts translate best to their authors.

As expected, the reflexive possessor in the above sentence cannot be bound by the dative.9 In
this example the dative is interpreted as the subject performing the action but the same result
is expected for affectedness datives. However, this is not what is demonstrated by the data. In
the sentence below, the dative is affected by the actions of an independent subject but the
reflexive possessor is nonetheless allowed within the dative constituent:

(18) [Swojej nauczycielce] Tomek nie zda egzaminu.


[his.refl. teacher]Dat. Tom.Nom. neg passed.3sg.m. exam.Gen.
Tom failed his exam and his teacher was affected by that.

The data introduced so far confirm that affectedness datives can be generated both in a high
and in a low position in the structure. Such availability of two attachment sites for the dative
is allowed provided that the dative does not simultaneously receive the interpretations
available in those two distinct positions.
The required evidence is provided by the combination of the reconstruction for binding
test and the interpretation of again. Namely, the dative constituent that contains the
reflexive possessor and therefore must have been generated below the subject position should
not be allowed to be interpreted outside the scope of again in sentences in which it occurs
sentence-initially, such as (19) below:

(19) [Swojej nauczycielce] ponownie Tomek nie zda egzaminu.


[his.refl. teacher]Dat. again Tom.Nom. neg passed.3sg.m. exam
Tom failed his exam again and his teacher was affected by that.

That this prediction is right is confirmed by the choice of scenarios compatible with the
above sentence. Namely, (19) can receive the interpretation paraphrased as Again, Tom
failing the exam affected his teacher but it cannot refer to a scenario The teacher was affected
(for the first time) by Tom again failing the exam. The unavailability of this interpretation is
expected as in order to bind the reflexive, the dative has to be interpreted in the scope of the

9
The same is true for the dative impersonal si construction:

(i) *Swoim dzieciom przyjemnie si taczyo.


[their.refl. children]Dat. pleasurably refl danced.3sg.n.
Intended meaning:*One=ones children danced and they experienced that as pleasurable.

However, in this example, even if the datives reconstruction to a lower position was possible, the agent
argument introduced by si would have to be identified with the dative in this case referring to the agents
children, which would create incomprehensible semantics for this structure.
UCLWPL 2012 130

adverb. However, the latter interpretation is available for the dative in structures in which it is
not modified by the reflexive possessor:

(20) Marysi ponownie Tomek nie zda egzaminu.


Mary.Dat. again Tom.Nom. neg passed.3sg.m. exam.Gen.
Tom failed his exam again and Mary was affected by that.

That means that the dative must be generated above the subject, providing evidence, just like
the sentence in (15), for the high attachment site of the affectedness dative.
The data provided in this section confirm that dative nominals associated with the agent
subject can only be generated in the highest position in the structure (cf. 17). However, both
high and low attachment sites are available for affectedness datives. There exists substantial
evidence that high dative DPs should not be analysed on a par with lower datives occurring in
sentences such as (10). However, due to space limitations in the present paper, I have only
been able to provide arguments for the above claim based on the reconstruction for binding
and the again test.

2 Basic analysis

As shown by (1) (4), dative DPs attaching in the highest position in the structure may
combine with a number of constructions that otherwise occur in the language as independent,
well formed structures. The relation between the dative nominal and the sentence that it co-
occurs with can in principle be expressed by the three representations provided below.

2.1 Semantic selection by the dative

The representations in (21) and (22) assume monoclausal structures in which the dative either
selects (21) or is selected by its sister constituent (22):

(21) TP

Dat TP (e)

(Adv) ... si ... V

The above configuration assumes that the dative directly selects its sister constituent, which is
a property associated with the adjunct status. In (22), the selection process proceeds in the
other direction: the dative is selected by the TP:

(22) TP

Dat TP (e)

(Adv) ... si ... V


UCLWPL 2012 131

Another possible configuration representing the relation between the dative and the co-
occurring sentence can be considered a special instance of (22). Namely, in the biclausal
structure in (23), the dative is also the selected element however, unlike in (22), it is not
introduced directly by its sister but by an additional syntactic head that establishes the
relation between the two constituents. The analysis of the dative as an element introduced by
such a head makes (23) a type of applicative construction parallel to Pylkknens High
Applicative in which the dative occupies the specifier position of a High Applicative Phrase
(XP) and an applicative head X is responsible for the semantics assigned to it.10

(23) XP

Dat

X TP (e)

(Adv) ... si ... V

In this representation a separate projection headed by X dominates the whole structure. The
head introduces a human dative DP in its specifier and includes a Tense Phrase (TP)
argument as an embedded clause.11 The Tense Phrase, depending on its structure, consists of
an impersonal si construction, a middle sentence, the anticausative or any other kind of
predicate.
In the following sections, I will present a number of arguments to support the claim that
the correct relation between the dative and the constituent it precedes is represented by the
structure in (21) in which the dative functions as a modifier.
The argument supporting semantic selection by the dative of the clause it attaches to is
based on the observation that regardless of the structure it occurs in, the dative requires that
the sentence it combines with expresses an event that the dative is associated with, but not in
control of. This implies that the semantics of the dative has the form in (24):

(24) [[Dat]] = xe [PARTICIPANT (x,e) & OUT OF CONTROL (x,e)]

10
An account along these lines has been proposed by Bosse, Bruening and Yamada (2011) for structures
occurring in a number of unrelated languages that allow a DP to be added to a clause with the interpretation that
that DP is psychologically affected by the event described by the verb phrase. Their proposal assumes that in
what they call the affected experiencer construction, the dative experiencer is introduced by a syntactic head
Aff(ect) that is also responsible for its affectedness semantics. However, Bosse et al. do not discuss any
structures in which the dative would be identified with the performer of the action.
11
An analysis based on the applicative model has also been proposed by Rivero, Arregui and Frckowiak
(2010) and Rivero and Arregui (2012), among others. In their proposal, the structure of Polish sentences such as
(1a) consists of an applicative phrase that is headed by a silent circumstantial modal. The ApplP takes the dative
as its specifier and the Tense Phrase and the adverb, both analysed as embedded clauses, are the arguments of
the modal. Whilst the analysis based on the presence of the modal head is supposed to account for the syntactic
and semantic properties of (1), due to the fixed position of the arguments of the modal with respect to one
another, the representation proposed by Rivero et al. faces certain difficulties with respect to, for instance, the
availability of interpretations that it predicts to be possible for sentences such as (1) that occur in structures
involving further modification. For example, when the structure in (1) interacts with negation, it can be
demonstrated that the interpretations that it receives differ from those predicted by Rivero et al.s proposal.
UCLWPL 2012 132

The above lack of control interpretation that characterises the dative occurring in (1) (4)
makes certain predictions with regards to the semantics that the dative can receive. In what
follows, I will assume that depending on the structure it occurs in, the datives lack of control
over the event can be expressed either when the dative is identified with the subject argument
of that event or when it is affected by that event. In order to be interpreted as the performer of
the action, the dative has to be co-indexed with the referent of the arbitrary subject argument
available in the structure that it precedes. As participants in a given event must be specified,
the requirement in (24) that the dative be interpreted as a participant in the event it is
associated with is satisfied when, as a result of co-indexation with the arbitrary subject
argument, the dative is linked to the subject thematic role. On the other hand, if no such
specification is provided, I will propose that through the application of the rule in (25),
participants are assigned a default affectedness interpretation.12 I will assume that this notion
of affectedness is interpreted as a presupposition.

(25) PARTICIPANT (x) AFFECTED (x)

Structures in which the dative cannot be identified with the external argument or an implicit
semantic subject include sentences in which the reflexive marker is present but it does not
mark the presence of the subject (e.g. the anticausative structure) and those that have a
separate logical subject performing or involved in the action (cf. 3). In those cases, the dative
receives the affectedness interpretation. Crucially, the dative can receive the affectedness
interpretation also when the reflexive marker identifies the presence of the external argument
or a semantic subject but they are not necessarily co-referent with the dative, as can happen in
the impersonal and in the middle structure.
As provided by (24), the inherent meaning of the dative is defined as out-of-control. I
will claim that the requirement that the category selected by the dative must be an event and,
consequently, the ungrammaticality of sentences such as (8), repeated as (26) below, in which
the dative combines with states, follows from this semantic description of the dative.

(26) *Jankowi lubi si Ani.


John.Dat. likes.3sg.n. refl Ann.Acc.
Intended reading: John likes Ann.

As out-of-control is part of the meaning assigned to subjects of subject experiencer verbs


such as lubi like, I will assume that the ungrammaticality of (26) in which the subject
experiencer verb is preceded by a DP also specified for the lack of control is caused by
economy conditions that block the derivation of this structure. In line with proposals
according to which a more costly derivation is permitted only when it produces a new
interpretation13, (26) is disallowed as its interpretation is the same as that of the nominative-
subject sentence in (27): both the dative and the nominative DP have no control over the state
expressed by the predicate.

(27) Janek lubi Ani.


John.Nom. like.3sg.m Ann.Acc.
John likes Ann.

12
There are other rules of this nature that assign default interpretations to categories, which remain
unspecified. See, for example, Harley and Ritter's (2002) discussion of person features.
13
See Reinhart (1995) for a discussion of phenomena affected by economy considerations.
UCLWPL 2012 133

In the case of agentive predicates, on the other hand, the use of dative DPs is motivated.
As the dative has no control over the action, in sentences in which it is identified with the
subject, it is interpreted as an involuntary agent/experiencer. As was already demonstrated by
(7), repeated as (28) below, the involuntary agent interpretation is also available for the
nominative-subject equivalent of the sentence in (1b):

(28) Janek kichn.


John.Nom. sneezed.3sg.m.
1.John sneezed (on purpose).
2.John sneezed (accidentally).

The availability of the involuntary agent interpretation for the nominative subject of the
above sentence is unexpected. Namely, as provided by the elsewhere principle, if the dative
case is available to encode that a given action was performed involuntarily, the fact that the
nominative DP is chosen over the dative should indicate that the agent was in control of that
action. However, as shown above, this is not the case in (28) as its subject could act either on
purpose or accidentally. The fact that there is no elsewhere effect in this case and that the
involuntary agent interpretation is available for (28) suggests therefore that (28) and (1) are
not in competition. This should not be considered surprising in the light of the proposal
introduced in this paper as sentences such as (1) are assumed to have a very different
structure to those represented by (28).

2.2 The arb rewriting rule

As was proposed earlier, depending on the structure that the dative combines with, its out-of-
control interpretation can be expressed either when it is identified with the subject participant
in an event or when it is affected by that event. However, both the performer of the action and
the affectedness interpretation are available for the dative that co-occurs with the middle or
the impersonal structure. In order to explain how these interpretations can be assigned to the
dative, I will follow Williamss (1980) proposal concerning the possibility of co-indexing
PRO subjects of infinitives marked as arb with DPs occurring in the sentence. Namely, I will
assume that in the impersonal si construction and in the middle structure the morpheme si
is assigned an arbitrary interpretation arb and that in both structures the subject agent marked
by the presence of siarb, whether syntactically active or only implicit in the semantics,
receives a generic interpretation. However, according to Williams, there are cases when it is
not possible to interpret arb generically. For example, the subject of the infinitive in the
following sentence has to be interpreted as John:

(29) It is important to John PROarb to leave.

The sentence above differs from the example in (30) below in which PRO can be controlled
either by John, Mary, or by neither of them:

(30) John told Mary that it would be important PROarb to leave early.

In the above examples arbs have to, as in (29), or may, as in (30), be subject to a nongeneric
interpretation. Williams proposes that they are rewritten as co-indexed with nominal phrases
UCLWPL 2012 134

in line with two rewriting rules14 which govern the assignment of antecedents to them.
According to these rules, any arb which is not rewritten is interpreted as generic or free in
reference. Also, as demonstrated by (29), arb rewriting rules are not subject to c-command.
In this example, the antecedent is within the PP but it is still able to provide a referent for the
pronoun.
Applying the above proposal to dative structures illustrated by (1) (4), I will assume
that the dative and the morpheme si in the impersonal si construction and in the middle
structure will have the same referent as a result of the application of a rule rewriting siarb as
co-indexed with the dative. The rule is formulated as follows15:

(31) arb i / DPDATi si_


Subject

As the arb rewriting rule is not obligatory, the reflexive morpheme will still be interpreted as
referring to a generic subject16 in sentences such as (32c) or (33c) below in which it is not co-
indexed with the dative. Therefore, in those cases it will receive the same interpretation as in
the impersonal structure in (32a) and in the middle construction in (33a) which are not
preceded by the dative:

(32) a Kicha siarb.


sneezes.3sg.n. refl
One sneezes.

b Jankowii sii kicha.


John.Dat. refl sneezes.3sg.n.
John (accidentally) sneezes.

c Jankowi siarb kicha.

14
See Williams (1980:216-17) for details.
15
For convenience, I will indicate that the rule identifies the dative with the morpheme si. In fact, the rule
co-indexes the dative with the implicit subject available in the structure that the dative precedes whose presence
is marked by si.
16
There are various factors that determine whether the dative becomes identified with the performer of the
action or whether it receives the affectedness interpretation. For example, in particular in the case of verbs
which describe actions typically controlled by the subject, the involuntary agent interpretation is more accessible
to the dative when the verb appears in the perfective rather than in the imperfective aspect:

(i) Jankowii sii zataczyo.


John.Dat. refl danced.3sg.n.perf.
John (accidentally) danced (once).

(ii) ??Jankowii sii taczyo.


John.Dat. refl danced.3sg.n.imperf.
John (accidentally) danced (repeatedly).

The above contrast is expected as the sentence in (ii) indicates that the dative performed the action more than
once which, in the absence of any additional information in the context, may seem pragmatically odd. On the
other hand, if the dative is to be interpreted as an affectee, this interpretation seems to be more accessible if the
action that affects the dative is performed repeatedly. Therefore, the affectedness interpretation in the context of
(i) and (ii) would be more compatible with the imperfective aspect of the verb.
UCLWPL 2012 135

John.Dat. refl sneezes.3sg.n.


One sneezes and John is affected by that.

(33) a Te buki siarb dobrze kroj.


these rolls.Nom. refl well cut.3pl.
These rolls cut well.

b Jankowii te buki sii dobrze kroj.


John.Dat. these rolls.Nom. refl well cut.3pl.
To John, these rolls cut well.

c Jankowi te buki siarb dobrze kroj.


John.Dat. these rolls.Nom. refl well cut.3pl.
These rolls cut well (to someone else) and John is affected by that.

The fact that the above rule affects both the impersonal and the middle structure suggests that
it applies either in the syntax semantics interface or only in the semantics. Namely, as in the
middle structure there is no syntactically active subject, in this case it is the implicit semantic
agent that is provided with a specific referent by the dative.
As demonstrated above, the arb rewriting rule is in operation in the context of the
middle and the impersonal si construction in which it determines the interpretation assigned
to the dative. However, there are also certain contexts where this rule does not apply. For
example, in the impersonal passive structure the dative can only receive the affectedness
interpretation which suggests that the arb rewriting rule which would identify the dative with
the referent of si is not in operation and, consequently, the pronoun receives a generic
interpretation:

(34) a *Jankowii bywa sii naladowanym.


John.Dat. is.3sg.n.habitual refl impersonated
Intended meaning: John is (regularly) impersonated.

b Jankowi bywa siarb naladowanym.


John.Dat. is.3sg.n.habitual refl impersonated
One is (regularly) impersonated and John is affected by that.

In the grammatical examples in (32), the interpretation of si as referring to the same


individual as the dative DP is licensed by its co-indexation with the dative which is the most
local element in the structure by which the pronoun can be provided reference.
As in the impersonal passive sentence in (34a) co-indexation is not allowed, there must
be a reason why the arb rewriting rule does not apply even though the locality condition
mentioned above seems to be satisfied. In line with what was said above about the arb
rewriting rule assigning reference to arbitrary subjects, I will assume that co-indexation of the
arb pronoun and the dative is disallowed in impersonal passive sentences because the
pronoun is interpreted as linked not to the external argument but to the logical object of the
passive verb. The difference between the two interpretations of si in active and passive
impersonal sentences can be illustrated by the two examples below:

(35) a Pobio si Janka (*przez Tomka).


beat-up.3sg.n. refl John.Acc. by Tom.Acc.
UCLWPL 2012 136

Intended reading: One beat John up (*by Tom).

b Byo si pobitym przez Tomka.


was.3sg.n. refl beaten-up by Tom.Acc.
One was beaten up by Tom.

As impersonal sentences such as (35a) contain a syntactically active subject, they cannot be
followed by the by-phrase. However, the by-phrase can be used in the passive impersonal
structure in (35b), which proves that si in this example is not linked to the presence of the
external argument but it indicates the arbitrary reference of the logical object of the passive
verb. Therefore, it is expected that in cases in which the referent of the object DP is a specific
individual, si will not be available in the structure. This is confirmed by the data:

(36) Janek by pobity przez Tomka.


John.Nom. was.3sg.m. beaten-up by Tom
John was beaten up by Tom.

As in (34) the referent of si is interpreted as the arbitrary logical object, it will not be co-
indexed with the dative given that the arb rewriting rule provides reference only for subject
arguments of structures that combine with the dative. If this is the case, there is no need to
refer to any locality conditions with regards to the DP that becomes co-indexed with the arb
pronoun.
The choice of one of the two interpretations available for the dative DP depends on
whether or not the dative is co-referent with the performer of the action. The fact that in the
passive structure only one interpretation is possible is an independent argument that supports
the existence of the arb rewriting rule operating in the context of dative sentences. In
addition, the difference in the acceptability of (34a) and (34b) confirms the claim put forward
in this paper that the affectedness and the actor interpretation of the dative are two separate
readings. If there was no real linguistic difference between those two readings, it would not
be possible for one interpretation to be available in the circumstances when the other was not
accepted.

2.3 Affectedness as a presupposition

According to the rule in (24), when the dative receives the actor interpretation, it is not in
control of the event in which it is a participant. In the affectedness scenario, the action is not
performed by the dative but there is a presupposition that the dative is psychologically
affected by that event. I will demonstrate that this claim is true on the basis of the Hey wait a
minute test applied to one of the affectedness dative structures.17 According to the test, in
response to the speakers utterance of Y, the hearer can reply Hey wait a minute I didnt know
X if and only if Y presupposes X. Therefore, in the context of (37a), when the hearer is
informed about Pauls actions, they can object along the lines of (37b):

(37) a Jankowi Pawe biega po swoim ogrodzie!

17
The same observation is also confirmed by the results of other tests typically used to identify
presuppositions. For example, the presupposition that John is affected by the event expressed by sentences such
as (37a) survives when this sentence is embedded under negation but is blocked in the reported speech context.
UCLWPL 2012 137

John.Dat. Paul.Nom. runs.3sg.m. around his.refl. garden

b Hej, zaraz, nie wiedziaem, e Janek tym


Hey, wait a minute, I didnt know that John.Nom. this.Instr.
Hey, wait a minute, I didnt know that John was affected by that!

si przejmuje!
worries.3sg.m.

On the other hand, the results of the Hey wait a minute test applied to datives which are
identified with actor arguments of their events confirm that there is no need to assume that
those datives are also affected:

(38) a Jankowi si zakaszlao!


John.Dat. refl coughed.3sg.n.
John (accidentally) coughed.

b Hej, zaraz, nie wiedziaem, e #Janek tym


Hey, wait a minute, I didnt know that John.Nom. this.Instr.
Hey, wait a minute, I didnt know that #John was affected by that!

si przejmuje!
worries.3sg.m.

The same results are obtained in the case of structures in which the dative is identified with
the subject and experiences the action in a particular manner:

(39) a Jankowi przyjemnie si taczyo.


John.Dat. pleasurably refl danced.3sg.n.
John danced and he experienced that as pleasurable.

b Hej, zaraz, nie wiedziaem, e #Janek tym


Hey, wait a minute, I didnt know that John.Nom. this.Instr.
Hey, wait a minute, I didnt know that #John was affected by that!

si przejmuje!
worries.3sg.m.

In that respect, the examples in (40) and (41) resemble nominative-subject sentences whose
agent subjects are not presupposed to be affected by the action that they perform:

(40) a Janek zakaszla.


John.Nom. coughed.3sg.m.
John coughed.

b Hej, zaraz, nie wiedziaem, e #Janek tym


Hey, wait a minute, I didnt know that John.Nom. this.Instr.
Hey, wait a minute, I didnt know that #John was affected by that!

si przejmuje!
UCLWPL 2012 138

worries.3sg.m.

(41) a Janek taczy z przyjemnoci.


John.Nom. danced.3sg.m. with pleasure
John danced with pleasure.

b Hej, zaraz, nie wiedziaem, e #Janek tym


Hey, wait a minute, I didnt know that John.Nom. this.Instr.
Hey, wait a minute, I didnt know that #John was affected by that!

si przejmuje!
worries.3sg.m.

The above data confirm that dative performers of the action receive the out-of-control
semantics but are not necessarily interpreted as affectees. Therefore, in this aspect they
pattern with nominative agents that are not expected to encode affectedness as part of their
semantics.
The proposal described above assumes that the affectedness interpretation for the dative
is only available when the dative is a participant in an event but in the structure it combines
with there is no argument to which it could be linked to receive the performer of the action
interpretation. It does not mean that an experiencer/involuntary agent dative cannot at the
same time be affected by the event that he performs. In fact, one would be expected to be
affected by a state that they experience or an action that they knowingly perform. In this case,
this additional affectedness interpretation would not be necessary for the individual in
question to become a participant in the event. However, if the dative is not the performer of
the action, as in the examples above, it can only be linked to that action via the notion of
affectedness.

2.4 Truth-conditional and non-truth-conditional meanings of dative constructions

As opposed to the out-of-control semantics that is part of the inherent meaning of the dative,
the affectedness interpretation is not linguistically encoded by the rule in (24). This may be
one of the reasons why affectedness datives contribute both truth-conditional and non-truth-
conditional meaning to the interpretation of the structure in which they occur. The existence
of the truth-conditional component of their meaning is confirmed by the fact that they can be
questioned and negated just like any other elements of the sentence carrying a lexical
meaning. For negation, this is demonstrated by the following data:

(42) a Nie jest tak, e Jankowi zepsuli samochd Tomka.


neg is so that John.Dat. broke.3pl. car.Acc. Tom.Gen.
It is not the case that John was affected by them breaking Toms car.

b Pawowi zepsuli samochd Tomka.


Paul.Dat. broke.3pl car.Acc. Tom.Gen.
Paul was affected by them breaking Toms car.

c Jankowi porysowali samochd Tomka.


John.Dat. scratched.3pl. car.Acc. Tom.Gen.
John was affected by them scratching Toms car.
UCLWPL 2012 139

As shown above, all constituents in the above examples can be negated. However, what
survives negation in these sentences is the notion of affectedness itself. For example, (42a)
above cannot mean that someone broke Toms car but that did not affect John. Similarly,
even if the answer to the question below is negative, it cannot refer to the lack of affectedness
on Johns part.

(43) Jankowi zepsu si samochd Pawa?


John.Dat. broke down.3sg.m. refl car.Nom. Paul.Gen.
Was John affected by Pauls car breaking down?

The above examples prove that the datives state of being affected by the event does not
affect the truth conditions of sentences in which the dative occurs. However, according to the
rule in (24), the inherent meaning of the dative is out-of-control. As this meaning
component is encoded in the syntax of the dative nominal, it should allow questioning and
negation. The examples in (44) and (45) confirm that this is indeed the case. In (44), the
answer contains a nominative instead of a dative DP which changes the involuntary action
into an intentional one. The correction context in (45) also involves replacing the dative with
the nominative DP to mark the agents control over the dancing:

(44) a Jankowi si zataczyo?


John.Dat. refl danced.3sg.n.
Did John accidentally dance?

b Nie, Janek zataczy.


neg John.Nom. danced.3sg.m.
No, John danced (intentionally).

(45) Nie jest tak, e Jankowi si zataczyo;


neg is so that John.Dat. refl danced.3sg.n.
It is not the case that John danced by accident;

Janek zataczy.
John.Nom danced.3sg.m.
John intentionally danced.

The data presented above provide evidence for the semantic specification of the dative
proposed in (24) which requires that the dative be a participant in the event. This requirement
is trivially satisfied if the dative controls si that is linked to an argument position. If this is
not the case, as a result of the application of the default rule in (25), the dative can be
interpreted as a participant even though it is not linked to a thematically marked argument.

3 Syntax of highest datives

The analysis proposed in this paper relies on the proposal that the relation between highest
datives and their sister constituents is represented by the structure in (46) in which the dative
nominal functions as an adjunct which c-commands and directly selects a constituent with a
particular semantic specification:
UCLWPL 2012 140

(46) TP

DPDAT TP (e)
Adjunct

(Adv) ... si ... V

The requirement that the clause selected by the dative DP describe an event as well as the
dependencies within the dative itself were defined by (24), repeated as (47) below, specifying
the inherent meaning of the dative:

(47) [[Dat]] = xe [PARTICIPANT (x,e) & OUT OF CONTROL (x,e)]

According to above rule, an individual (with an appropriate [+sentient] semantics) is selected


by the dative marker and it receives the out-of-control interpretation, as illustrated by (48). I
will assume that the dative is a functional category which does not prevent the argumental
properties of the DP that it selects from percolating to the higher node. As a result, the dative
DP combines both adjunct and argument properties.

(48)
TP

DP TP(e)
Adjunct

DP DAT
Argument

The above representation makes certain predictions with regards to the status of the dative
DP. The initial evidence in favour of the adjunct status of dative DPs in (1) (4) is based on
their three important properties. Namely, in all of the examples above, the dative nominal is
an optional element of the structure, it makes a constant semantic contribution to the meaning
of the constituent that it modifies and it seems to select its sister constituent on the basis of its
semantic specification. Additional arguments in favour of (48) are provided by binding and
secondary predication data.18

18
There is some evidence to suggest that mixed adjunct and argument properties of the dative are also
illustrated by their weak islands extraction patterns. However, in order to claim that this is the case, it is
necessary to collect more data illustrating weak islands extraction properties of dative DPs that can be
interpreted both as involuntary agents/ experiencers and as affectees.
UCLWPL 2012 141

3.1 Optionality and semantic selection by the dative

As far as the optionality argument is concerned, the semantic or syntactic well-formedness of


the above sentences does not depend on the dative DP. Therefore, in the absence of a dative
nominal, in line with the semantics of the middle structure, by whoever utters the sentence in
(4a), the translation task described by that sentence will be considered easy. Similarly, (1a)
used without the dative will also remain grammatical. It will become an impersonal sentence
referring to the action of dancing performed by an arbitrary subject who will perceive that
action as pleasurable.
Another difference between adjuncts and arguments concerns the fact that adjuncts are
systematically optional whilst arguments can be left out from the structure only under certain
conditions. As was indicated earlier, the dative is never an obligatory constituent of the
structures illustrated by (1) (4). This is a property characteristic of adjuncts. Arguments, on
the other hand, can only be left over in some specific cases. For example, objects of some
transitive verbs can only be omitted if the interpretation of the verb object unit is specified
in the lexicon and, as a result, its reading remains the same even when the object has been left
over. This happens in the case of verbs such as smoke or drink that typically refer to smoking
tobacco and drinking alcohol also when their objects are not explicitly present. Expression of
a direct object is sometimes optional also in the case of some other transitive verbs, e.g. read,
win, bake, etc. However, alongside such verbs there are those that never allow their objects to
be omitted. In that respect, the distribution of the dative is much more regular than that of
direct objects. Namely, the dative in sentences illustrated in 1 - 4 is never an obligatory
element of the structure and as such it can be freely omitted without having to satisfy any
semantic or syntactic requirements.
Also, the examples provided in (5a) and (5b) cannot be treated as instances of (1a) and
(1b) in which the dative would have been left out according to the same principles that allow
the omission of pragmatically inferable argument pronouns in a number of pro-drop
languages. As demonstrated by the two examples below, only (49b) is an acceptable
continuation of (49a). The unacceptability of (49c) in this context confirms that this sentence
is an impersonal structure referring to an action carried out by an arbitrary subject and as such
it does not contain a phonologically null dative pronoun:

(49) a Czego chciae si dowiedzie o Janku?


What did you want to find out about Janek?

b e wanie wyszed?
that just left.3sg.m.
That he has just left?

c #e przyjemnie si wczoraj taczyo?


that pleasurably refl. yesterday danced.3.sg.n.
Intended meaning: That yesterday he danced with pleasure?

The above data confirm that datives pattern with adjuncts rather than with arguments. Just
like adjuncts, they are always optional. Even though in some circumstances arguments can
also be considered optional, there exist certain conditions specifying when they can be left
out from the structure. There are no similar restrictions in the case of the dative.
UCLWPL 2012 142

Arguments and adjuncts differ also in terms of the way in which they semantically
contribute to the meaning of the sentence they occur in. Whilst arguments demonstrate a lot
of variation in terms of their semantic interpretation, it has typically been assumed that
adjuncts make a constant semantic contribution to the meaning of the constituent that they
modify.19 Therefore, even though adjuncts belong to different groups distinguished on the
basis of their meanings, a particular adjunct will always contribute to the meaning of a
sentence in the same way. For example, adverbs of time will always specify the temporal
frame of the modified event regardless of the other properties of the structure in which they
occur. Arguments, on the other hand, will receive a different interpretation depending on the
kind of syntactic head they are selected by.20 As a result, even the interpretation assigned to
arguments that constantly occur in the subject or the object position in a clause will vary
among agents, experiencers, themes, etc. For instance, the DPs in run a mile, knit a sweater
or worry John will be interpreted in a different way even though they occupy the same
position within their verbal phrases. The fact that unselected datives always receive the same
interpretation regardless of the structure that they are followed by demonstrates that in this
respect they also pattern with adjuncts.

3.2 Binding

One of the assumptions of the traditional binding theory is that binding is only possible from
an argument position. However, this claim cannot be true for pronominal binding as a
pronoun can be bound by a quantificational expression occupying an A-bar position in
sentences such as (50) below21:

(50) Every dayi John thinks iti is the best day of his life.

(50) can be accounted for if it is assumed that pronominal binding from an adjunct position
should be allowed as long as it does not create weak cross-over effects. If this is the correct
proposal, subject to locality conditions on binding, the dative should be able to bind
pronominal elements occurring in the clause it combines with. On the other hand, anaphoric

19
A proposal trying to undermine this claim could potentially rely on the observation that the adverb
cleverly in the following examples can either describe the way in which the car was repaired (i) or it can refer
to the fact that the car was repaired at all (ii), which would suggest that the meaning of the adverb changes
depending on the position in which it attaches:

(i) John has repaired the car cleverly.


(ii) John cleverly has repaired the car.

However, the above data can also be explained in the following way. One possibility is that there are two types
of cleverly and each of these has unique semantic contribution. According to the other solution, there is only
one cleverly that has the same semantic contribution but its interpretation depends on what element it
combines with. Crucially, regardless of which of the two options is chosen, the adverb still makes single
semantic contribution to the meaning of the constituent that it modifies and the claim that its interpretation
depends on its attachment site can no longer be maintained.
20
Proposals that would like to relate those different interpretations to different syntactic structures could
rely on data showing that cross-linguistically certain types of objects could occupy a different position from
other objects with respect to the head that they are selected by. However, there has been no evidence available
as yet to confirm that such data actually exist.
21
Example adopted from Adolfo Ausns abstract on Pronoun binding and the A/A-bar distinction. The
26th University of Pennsylvania Linguistics Colloquium. 2003.
UCLWPL 2012 143

binding by the dative is expected to be impossible.22 These predictions are confirmed by the
data. The sentences in (51) and (52) below are ungrammatical due to principle B violation
but, as expected, long distance binding by the dative of a pronominal subject (53) and object
(54) is possible:

(51) *Kademu chopcui proi uderzy Pawa.


[every boy]Dat. pro hit.3sg.m. Paul.Acc.
Intended meaning: Every boy hit Paul and they were affected by that.

(52) *Kademu chopcui Pawe goi uderzy.


[every boy]Dat. Paul.Nom. he.Acc. hit.3sg.m.
Intended meaning: Paul hit every boy and they were affected by that.

(53) Kadej nauczycielcei Pawe obla egzamin bo proi


[every teacher]Dat. Paul.Nom. failed.3sg.m. exam.Acc. because pro
Every teacher was affected by Paul failing the exam because they had badly

go niedobrze przygotowaa.
he.Acc. badly prepared.3sg.f.
prepared him for it.

(54) Kadej nauczycielcei Pawe obla egzamin bo jeji


[every teacher]Dat. Paul.Nom. failed.3sg.m. exam.Acc. because she.Gen.
Every teacher was affected by Pawel failing the exam because he didnt listen to

nie sucha.
neg listened.3sg.m.
them.

The above sentences demonstrate that even though the dative is an adjunct, it is allowed to
bind pronominal elements occurring in the clause it precedes. As illustrated by (55) and (56)
below, that includes binding pronominal possessors contained within the subject or the object
constituent23 of the clause:

22
There are no available data to suggest that anaphoric binding from an adjunct position is allowed. The
sentence in (i) that contains a reflexive pronoun is clearly ungrammatical:

(i) *Every dayi John admires itselfi.


23
The grammaticality of (55) and (56) indicates that pronominal possessors in Polish are exempt from
principle B effects and allow local binding. If this is the case, the fact that the pronoun cannot be locally bound
by the subject in sentences such as (i) below should not be considered the result of principle B violation. Instead,
as there is a subject-oriented anaphor in Polish, it is preferred over the pronoun:

(i) Pawei poda Jankowij swoji,*j/ jego*i,j ksik.


Paul.Nom. passed.3sg.m. John.Dat. his.refl. his.pron. book.Acc.
Paul passed his (Pauls) book to John.
UCLWPL 2012 144

(55) Kademu piekarzowii jegoi chleb si dobrze kroi.


[every baker]Dat. his.pron. bread.Nom. refl well cuts.3sg.m.
To every baker his bread cuts well.

(56) Kademu piekarzowii klienci chwal jegoi chleb.


[every baker] customers.Nom. praise.3pl. his.pron. bread.Acc.
Every bakerDat. is affected by customers praising his bread.

On the other hand, as anaphoric binding is only possible from an argument position, it should
not be possible for reflexive anaphors or reciprocals to be bound by the dative. However, due
to their subject-orientation24, reflexive anaphors cannot be used to argue that anaphoric
binding by the dative is disallowed. Reciprocals, on the contrary, can be bound by non-
subject constituents and therefore they can provide evidence in favour of the argument or the
adjunct status of the dative. As demonstrated by (57) below, the reciprocal expression sobie
nawzajem self reciprocally can be bound by the subject and by the object constituent but,
crucially, not by the dative:

(57) Obydwu nauczycielomi [Pawe i Tomek]j przedstawili


[both teachers]Dat. [Paul and Tom]Nom. introduced.3pl.vir.
Both teachers were affected by Paul and Tom introducing the girls to

dziewczynyk sobie*i,j,k nawzajem.


girls.Acc. self.Dat. reciprocally
each other.

As was demonstrated above by sentences such as (51) and (52) that give rise to principle B
effects, the dative must belong to the binding domain of the clause that it precedes. If this is
the case, the fact that binding of the reciprocal expression by the dative in (57) above is not
available can only be accounted for if it is assumed that the dative occupies an A-bar position
from which anaphoric binding is not allowed.
The above structures confirm that the dative can bind pronominal but not anaphoric
elements occurring in the clause that it combines with. This observation is also true in cases
when the dative precedes the impersonal structure represented by (58) below in which x
stands for an anaphoric or pronominal object or possessor:

24
The two sentences below demonstrate that the reflexive sobie self can only be bound by the subject:

(i) Janeki kaza Tomkowij kupi sobiei,j samochd.


John.Nom. told.3sg.m. Tom.Dat. buy.Inf. self.Dat. car.Acc.
John told Tom to buy a car for himself.

(ii) Janeki obieca Tomkowij kupi sobiei*j samochd.


John.Nom. promised.3sg.m. Tom.Dat. buy.Inf. self.Dat. car.Acc.
John promised Tom to buy a car for himself.

In (i), the reflexive can be bound by the local or by the matrix subject. Crucially, as in (ii) the object cannot act
as a binder, binding of reflexive anaphors cannot be used as a diagnostic to determine the status of the dative.
UCLWPL 2012 145

(58) DAT si x

As was proposed before, the choice of one of the two interpretations available for the dative
DP depends on whether or not the dative is the performer of the action. According to the arb
rewriting rule introduced in the previous section, the dative is associated with the subject
argument when it provides reference for the arbitrary pronoun. When the rule does not apply,
si is interpreted as an arbitrary subject different from the dative. Therefore, when it comes to
anaphoric/ pronominal binding, sentences with siarb functioning as the subject do not differ
from cases such as (52) or (56) where the subject is a specific individual different from the
dative and the dative receives the affectedness interpretation:

(59) Kademu chopcu i uderzao siarbj siebie*i,j/ go*i,* j,k.


[every boy]Dat. hit.3sg.n.imperf. refl self/ he.Acc.
Intended meaning: Every boy was affected by one/people hitting themselves.

As in the sentence in (52), principle B prevents the dative in (59) from binding the
pronominal object go him. The pronoun cannot be bound by the reflexive subject either as
the subject locally can only bind reflexive anaphors. Similar patterns can be observed for
pronominal and anaphoric possessors. As in the above sentence, the reflexive anaphor in (60)
can only be bound by the subject. In (61), the pronominal possessor cannot be locally bound
by the subject and as it does not give rise to principle B effects, it can be bound by the dative:

(60) Kademu chopcui biegao siArbj po swoim*i,j ogrodzie.


[every boy].Dat. ran.3sg.n.imperf. refl around his.refl. garden
Intended meaning: Every boy was affected by one/people running in their (ones)
garden.

(61) Kademu chopcui biegao siArbj po jegoi,*j,k ogrodzie.


[every boy].Dat. ran.3sg.n.imperf. refl around his.refl. garden
Intended meaning: Every boy was affected by one/people running in their (every
boys) garden.

The above predictions with regards to the possibility of anaphoric/ pronominal binding by the
dative are also confirmed when as a result of the application of the arb rewriting rule the
dative is associated with the impersonal subject and interpreted as the performer of the action,
as in the sentence in (62):

(62) Jankowii przyjemnie sii biegao.


John.Dat. pleasurably refl ran.3sg.n.imperf.
John ran and he considered that pleasurable.

If the impersonal clause selected by the dative contains a possessor pronoun referring to the
dative, it is expected that the possessor will have to be expressed by the reflexive swj. This
is confirmed by the data:

(63) Jankowii przyjemnie sii biegao w swoimi ogrodzie.


John.Dat. pleasurably refl ran.3sg.n.imperf. in his.refl. garden
UCLWPL 2012 146

John ran in his garden and he considered that pleasurable.

As subjects cannot locally bind pronominal elements in Polish, it is expected that the above
sentence should not remain grammatical when swj is replaced by the pronominal possessor
jego. However, such data can be accounted for if it is assumed that the pronominal is
interpreted as co-referent with and not bound by the dative.25
As demonstrated above, the data provided so far support the configuration in (48).
Further evidence to confirm that this proposal correctly represents the relation between the
dative and the constituent that it selects comes from structures that illustrate the effects of
binding principle C. Namely, as predicted by (48), the sentence below in which the dative
pronoun is co-indexed with the R-expression contained in the clause that it precedes is
ungrammatical:

(64) *Jemui Marysia uderzya Pawai.


he[strong]Dat. Mary.Nom. hit.3sg.f. Paul.Acc.
Intended meaning: Mary hit Paul and he was affected by that.

If the above example is excluded by principle C, it should be possible to construct sentences


in which a possessor contained within the dative constituent could be co-indexed with an R-
expression. The following data demonstrate that such sentences are indeed grammatical. As
the pronominal possessor in (65) does not c-command the R-expression, this sentence does
not give rise to principle C effects.

(65) Jego mamiei Marysia uderzya Pawai.


[his mum]Dat. Mary.Nom. hit.3sg.f. Paul.Acc.
Intended meaning: Mary hit Paul and his mum was affected by that.

The sentences in (64) (65) demonstrate that condition C effects in dative structures are
consistent with the proposal in (48). Even though a detailed study of how principle C operates
in Polish would be necessary to draw further conclusions with regards to the data, on the
basis of the above examples it seems that the analysis proposed in this paper is likely to make
the right predictions in this respect.
Summing up, the totality of the binding data presented above are easily accounted for if
it is assumed that the dative in (1) (4) should be analysed as an adjunct that is generated in
the highest position in the structure from which it c-commands the clause that it combines
with.

25
This can be checked with the use of the data that illustrate the availability of a bound reading for
sentences involving the operator tylko only that contain the reflexive or the pronominal possessor:

(i) Tylko Jankowi dobrze spao si w swoim/ jego ku.


only John.Dat. well slept.3sg.n. refl in his.refl/ pron. bed
Only John slept well in his bed.

In order to express the meaning Only Janek is an x such that x slept well in xs bed, speakers of Polish prefer to
use the version of (i) that contains the reflexive possessor. On the other hand, in order to say Only Janek is an x
such that x slept well in Janeks bed, the pronominal possessor is chosen over the reflexive one. This suggests
that the relation between the dative and the pronominal possessor is co-reference and not binding.
UCLWPL 2012 147

3.3 Secondary predication

As secondary predicate licensing conditions include the requirements that a secondary


predicate has to be c-commanded by its subject and that the subject has to occupy an A-
position, secondary predication data can provide a further argument to support the
configuration in (48).
The role of the c-command restriction on predication is demonstrated by the following
examples:

(66) a John ate the dish warm.


b * John ate from the dish warm.

In both cases the subject of the secondary predicate occupies an argument position. However,
in the ungrammatical (66b), the dish does not c-command the adjective that it is supposed to
be modified by because it is contained inside a PP which does not contain warm; in (66a)
there is c-command from the DP to the modifying adjective.
On the contrary, in the case of the structure in (67), the c-command restriction on secondary
predication is the only requirement that is satisfied. Due to its adjunct status, the dative DP
cannot be modified by a secondary predicate26:

(67) * Jankowi Pawe biega po ogrodzie nago.


John.Dat. Paul.Nom. runs.3sg.m. around garden naked
Intended meaning: Paul runs in the garden when John is naked and John is
affected by that.

As expected, the above sentence becomes acceptable when the secondary predicate refers to
the nominative subject of the clause preceded by the dative:

(68) Jankowi Pawe biega po ogrodzie nago.


John.Dat. Paul.Nom. runs.3sg.m. around garden naked
Intended meaning: Paul runs naked in the garden and John is affected by that.

Similarly, the fact that the dative does not occupy an A-position accounts for the
ungrammaticality of the middle (69) and the anticausative (70) structure below:

(69) * Jankowi te teksty najlepiej si tumacz nago.

26
The fact that the dative can be the subject of the secondary predicate in sentences such as (i) below is
due to the fact that it is identified with the subject of the impersonal sentence that occupies an A-position and c-
commands the secondary predicate that it is modified by:

(i) Jankowii najlepiej czyta sii t ksik nago.


John.Dat. best read.3sg.n. refl this book naked
John reads this book and he enjoys the reading most when he reads it naked.
UCLWPL 2012 148

John.Dat. these texts.Nom. best refl translate.3pl. naked


Intended meaning: To John these texts translate best when John is naked.

(70) * Jankowi kawa wylaa si nago.


John.Dat. coffee.Nom. spilt.3sg.f. refl naked
Intended meaning: The coffee spilt and John was affected by that when he
was naked.

3.4 The status of the dative summary

The data provided in this section confirm that the relation between the dative DP and the
constituent that it selects in structures illustrated by (1) (4) is correctly represented by (48).
According to the rule in (24), the dative DP has the adjunct status with regards to the clause it
combines with but within the adjunct constituent, the dative marker selects a noun phrase and
provides it with the required semantics. These properties have been illustrated by numerous
binding and secondary predication data. Further arguments in favour of (48) can possibly be
provided by weak island extraction patterns but more research needs to be carried out to
collect the relevant data.

4 Summary and alternative proposals

The data presented in the previous sections illustrate a number of syntactic structures that can
be preceded by a dative DP. They include the impersonal si construction, the middle and the
anticausative structure and various types of other intransitive and transitive sentences.
The variety of structures that can combine with the dative can also be argued to provide
further evidence for the representation in (21) and the proposal that the dative selects the
structure that it co-occurs with and, consequently, against the structure in (22) in which the
selection process happens the other way round. Namely, the proposal in (22) is undermined
by the fact that there are no semantic or syntactic properties that the above mentioned
constructions could have in common to form a uniform category that would select and theta-
mark the dative argument. On the contrary, as according to (21), if the dative does not have
the syntactic status of an argument, it can attach to a category in which all thematic roles have
already been satisfied. This is characteristic of a process involving adjunct modification.
Also, additional evidence against the representation in (22) concerns the fact that the presence
of the dative in (1) (4) is optional. As according to (22) the dative is selected by the clause
that it precedes, in order for this proposal to be correct, the dative would have to be an
obligatory element of the structure.
The main alternative to the proposal put forward in this paper has already been
illustrated by the representation in (23), repeated as (71) below. It involves a structure in
which the dative is a constituent introduced by an additional syntactic head that establishes a
relation between the dative and the event expressed by the verb:

(71) XP

Dat

X TP (e)

(Adv) ... si ... V


UCLWPL 2012 149

Regardless of the semantics of X, the above structure cannot correctly predict the meaning of
the dative, its properties and the properties of structures in which the dative occurs. If X was
a functional head in the extended projection of the verb, it would be difficult to find
phonological evidence confirming its presence due to the fact that X would not be
phonologically realised. Also, because of the word order in dative-nominal sentences in
which the VP normally follows the dative, there is no clear syntactic evidence for the
existence of such a functional projection. In addition, in the absence of arguments supporting
the presence of a functional head X, there is some evidence arguing against its existence.
Namely, if X was a functional head in the extended projection of the verb, it would be
expected to constitute an obligatory element of the structure. However, that this cannot be the
case is demonstrated by the fact that the presence of the dative, the constituent supposed to be
introduced by X, is optional in all structures discussed in this paper.
Alternatively, if X was a lexical head, it would have its own event structure. An
analysis of structures in (1) (4) involving the dative being introduced by a head with this
semantics would be inspired by Pylkknens analysis of applicative arguments in which the
highest dative would be introduced by a high applicative head denoting a relation between the
applied argument and the event expressed by the sentence it combines with. It could be
possible to find evidence for the presence of a lexical head introducing the dative by looking
at adverbial modification of dative-nominal sentences. Namely, if the datives state of
affectedness could be modified by a different temporal adverb than the event that caused that
state, that would constitute evidence for the presence of an additional lexical head introducing
the dative and assigning the affectedness semantics to it. However, as demonstrated below,
sentences containing a highest dative DP do not accept modification by time adverbials
referring to different time frames:

(72) * Jankowi we wtorek zepsu si samochd Tomka.


John.Dat. in Tuesday broke down.3sg.m. refl car.Nom. Tom.Gen.
Intended meaning: John was affected on Tuesday by Toms car breaking down at

w weekend.
in weekend
the weekend.

The above sentence containing an affectedness dative is ungrammatical even though it is not
uncommon to experience the effects of an event some time after the event occurred.
Therefore, (72) cannot be used to provide motivation for the existence of an additional
projection responsible for the affectedness semantics.
The same conclusion can be drawn on the basis of dative-nominal sentences that
contain adverbials sensitive to aspectual properties of the modified VP. For example, a
durative adverbial can only occur in a sentence containing an atelic event predicate. In the
two sentences below, the verb can be interpreted as atelic only in (74) as its imperfective
aspect allows a multiple-event reading:

(73) Jankowi (*przez tydzie) zepsu si samochd Tomka


John.Dat. for week broke down.3sg.perf. refl car.Nom. Tom.Gen
*John was affected by Toms car breaking down once for a week.

(74) Jankowi przez tydzie psu si samochd Tomka


UCLWPL 2012 150

John.Dat. for week broke down.3sg.imperf. refl car.Nom. Tom.Gen


John was affected by Toms car breaking down repeatedly for a week.

If for a week could be introduced into the sentence in (73), that would confirm that this
structure contained a separate verbal projection encoding the affectedness semantics of the
dative just like the stative expression odczuwa skutki czego be affected by the effects of
sth:

(75) Janek przez tydzie odczuwa skutki tego, e


John.Nom. for week felt.3sg.m.imperf. effects this.Gen. that
John was affected for a week by Toms car breaking down.

zepsu si samochd Tomka.


broke down.3.sg.perf. refl car.Nom. Tom.Gen.

As described above, in order to confirm the presence of an applicative head in the relevant
structures, it would be necessary for sentences such as (72) and (73) to be grammatical.
Therefore, the fact that they cannot be modified by two independent time adverbials means
that there is no positive evidence to confirm the existence of an additional projection
introducing the notion of affectedness into the relevant dative structures.

5 Conclusion

In this paper I have proposed that highest dative DPs in (1) (4) should be analysed as
adjuncts with the inherent out-of-control semantics defined by (24) and an internal syntactic
structure represented by (48). This analysis correctly predicts the meaning of structures in
which the dative occurs and it accounts for the two interpretations that can be assigned to the
dative in relevant contexts. This cannot be achieved by proposals based on the applicative
model that have been assumed to account for the properties of at least some non-argumental
dative structures in other languages.

References
Ausn, A. (2003). Pronoun binding and the A/A-bar distinction. Abstract for The 26th University of
Pennsylvania Linguistics Colloquium.
Bosse, S., Bruening, B., & Yamada, M. (2012). Affected experiencers. Natural Language and Linguistic
Theory, 30, 11851230.
Harley, H., & Ritter, E. (2002). Person and Number in Pronouns: A Feature-Geometric Analysis. Language, 78,
482526.
Kibort, A. (2004). Passive and passive-like constructions in English and Polish. Ms., Cambridge: University of
Cambridge.
Pylkknen, L. (2008). Introducing arguments. Cambridge, Mass: MIT Press.
Reinhart, T. (1995). Interface strategies. OTS Working Papers in Linguistics, Utrecht University.
Rivero, M., Arregui, A., & Frckowiak E. (2010). Variation in circumstantial modality: Polish vs. Sttimcets.
Linguistic Inquiry, 41, 704-714.
Rivero, M., Arregui, A., & Frckowiak E. (2012). Building Involuntary States in Slavic. In V. Demonte & L.
McNally (Eds.), Telicity, change, and state: A cross-categorial view of event structure (pp 300-332).
Oxford University Press, New York.
Williams, E. (1980). Predication. Linguistic Inquiry, 11, 203-38.

You might also like