Involuntary state constructions in Polish: a case of semantic-syntactic interface

Pobieranie 133.28 Kb.
Rozmiar133.28 Kb.

Involuntary state constructions in Polish: a case of semantic-syntactic interface

Anna Malicka-Kleparska

1. Introduction

Interfaces in language can be understood in at least two ways: they may be looked for in the linguistic material, where certain structures are mixtures of e.g. phrasal and lexical elements, or lexical and sub-lexical ones (and such examples abound in Polish), or, alternately, they can be traced in the integration and co-functioning of traditional grammatical subcomponents (e.g. semantics and syntax), as is the case in current approaches to semantic-syntactic investigations within generative grammar, fuelled by works by e.g. Marantz (1984), Kratzer (1996), Heim and Katzer (1998), Pylkkänen (2008) etc. We have selected both of these options to talk about interfaces, and thus we will take up selected argument structure/valency phenomena in Polish and analyse them in the light of what we view as major insights of Heim and Kratzer’s (1998) Semantics in Generative Grammar.1

To substantiate the above pronouncements, let us give some indication as to the type of material and type of theory that we have in mind. Certain transitive and anticausative2 structures in Polish can serve as the simplest example of the interfaces. Here the same meaning can be expressed with the use of either the standard syntactic structure: SVO ­­– Oni spakowali rzeczy ‘They have packed up their things’, or with the use of a lexically expressed anticausative predicate, e.g. pakować się ‘pack up things’, incorporating in its meaning the notion of the object – Oni spakowali się ‘They have packed up their things’. Thus, the sentence with both the complex predicate and the overt object is ungrammatical: *Oni spakowali się rzeczy. The first sentence contains the full-fledged ingredients of a syntactic structure, all clearly realised as lexical items, while in the second the object is covert. This transformation affects the form of the linguistic expression and it is morphologically marked by the clitic, pseudo-reflexive element się, which in itself belongs to the interface of morphology and phonology, as it is semi-independent of the lexical verb.3 The semantic properties of the structure do not change along with the formal reshufflings, contrary to our expectations in such cases.

Of course the opposite situations also occur, i.e. there are such cases where, at first glance, a change of meaning does not require a structural change. For instance, the sentence Piotr pakuje may equally well stand for ‘Peter is packing things up’, where two arguments participate in the situation (Agent – Peter, Theme – Things), and for ‘Peter is working out’ with a single participant (Agent). In this paper we will consider the problems of matching most efficiently semantic and morphosyntactic representations, so that the interfaces are expedient for a very problematic group of Involuntary-state constructions (ISCs) in Polish.

The framework we have chosen seems to be ideal for dealing with such interfaces, as it is generative semantics in the tradition of Heim and Kratzer (1998), where no distinction is drawn between intra- and inter-lexical levels and where semantic structures freely permeate syntactic and lexical entities. In other words, for a semantic interpretation it does not matter whether on the plain of form we are dealing with a lexical item, a bound morpheme or a phonologically unrealised formal marker.

2. Theoretical background (Semantics in Generative Grammar, Heim and Kratzer 1998)

Most seminal work in the interface area between syntactic structures and semantic investigations seems to have been done within Heim and Kratzer’s tradition of Semantics in Generative Grammar (1998). In this approach, the essential role is played by denotations, which have lexical representations and these, in turn, are combined into complex structures according to certain principles. Denotations may represent individuals, truth values, and functions from individuals to truth values. Then, in the lexicon we have denotations (also called extensions) of elements, e.g. (p. 15) [[Ann]], which stands for an individual called Ann, 1 standing for true and 0 for false, and functions, which are represented by verbs (among others). For instance, the denotation: [[smokes]] equals the function f: the set D (individuals, e.g. Ann) → to {0, 1}, the set of truth values, for all such x’s, where x belongs to D, f(x) is true (equals 1) if and only if x smokes.

The other sub-system in this semantic component consists of three general principles (Heim and Kratzer 1998: 43–44):

  • Terminal Nodes (TN)

If α is a terminal node, [[α]] is specified in the lexicon.

This should be interpreted in the way that each terminal node in a structure must have its own lexical representation.4

  • Non-Branching Nodes (NN)

If α is a non-branching node, and β is its daughter node, then [[α]] = [[β]].

That is, the non-branching node denotes the same as its terminal lexical representation.

  • Functional Application (FA)

If α is a branching node, [ γ] is the set of α’s daughters, and [[β]] is a function whose domain contains [[ γ]], then [[α]] = [[β]]([[ γ]])

This principle reads that if we are dealing with a branching node, one daughter stands for the function from an argument to its truth value, and the other for the argument belonging to the domain of this function. A further consequence of the above system is that language structures are predicted to be, at most, binary. In other words, all structures that might look different are in fact reducible to binary divisions. For instance, a structure with a transitive verb might look as if it were not binary, as it contains the subject, the verbal element and the object. However, in this system it will be perceived as consisting of two binary levels of one hierarchy. For instance (p. 27), a transitive verb, as in x likes y, has the denotation of two functions, e.g. f and g. F is a function from individuals in a given domain to function g, while g is a function from individuals in the same domain to truth values, whose truth value is 1 iff x likes y.

This view of semantic representation is in agreement with the assumption that the external argument and its VP constitute one layer of a sentence structure, while the internal argument is in a closer commune with the verbal node (see Marantz 1984), and both layers are strictly binary. Of course, it is not possible to present the theory even in a most general way within the limits of this paper, but the structures and notations we will be using and quoting here are rooted in this tradition.

The theory will be seen at work below when we consider the concrete linguistic material which consists of the so called ISCs (Involuntary-state constructions) in Polish. A reference point will be supplied by an analysis of these constructions as presented in Rivero, Arregui and Frąckowiak (2010).

3. Introducing the linguistic material: ISCs (Involuntary-state constructions) in Polish
Inoluntary-state constructions of the type analysed in Rivero, Arregui and Frąckowiak (2010: 704–705) constitute a very interesting and semantically varied group of structures in Polish:

(1) Marta chciała zjeść ciastko, a jej się kichnęło.

Marta wanted to eat a cookie but she.DAT REFL sneezed.NEU

‘Marta wanted to eat a cookie, but she suddenly had to sneeze.’

(2) Napisało mi się własne imię.

wrote.NEU I.DAT REFL own name.ACC

‘I wrote my own name (by chance).’

The form of these structures is characterised by the presence of an argument in the Dative case, whose function is established with respect to the event specified by the verbal element. In the example in (1) the dative argument is something like the Theme of the event, the involuntary undergoer of the activity. In example (2) the dative argument resembles an involuntary agent of the activity, where by chance refers to the fact of writing one’s own name. Yet another different interpretation is connected with examples like (3) below:

(3) Tańczyło mi się dobrze.

danced I.DAT REFL well

‘I danced well and the dancing was good.’

Here the dative argument clearly represents the Agent of the activity, and the by chance interpretation present in (1) and (2) is by no means obligatory in (3), as we may have sentences like:

(4) Tańczyło mi się dobrze, ponieważ dobrze odpoczęłam.

‘I danced well, as I had taken a rest.’

Other points of interest include the appearance of the verb in the singular neuter, the form that in these examples shows no predilection for agreeing with any of the arguments present in the structure. Such singular neuter predicates may constitute default, elsewhere, verbal forms5 in Polish, frequently used for impersonal6 constructions, though we would like to stress here that they are not identical with infinitives (a fact which we will show to be vital in section 5.1.3). Neuter verbs also appear in Polish in another kind of impersonal constructions, created with bi-argumental verbs of the canonical Causer verb Theme structure (with well understood Causer, and well seen effect exerted on the Theme), e.g. (from Laskowski 1984: 147):

(5) Złamało mu nogę

broke.NEU he.DAT leg.ACC

‘His leg got broken (by sth understood).’7

Consequently, the 3rd person neuter singular form of the verb may be treated as something outstanding in the system of regular inflected verbs in Polish, as it performs a number of jobs.

Yet another interesting feature of ISCs (as claimed by Rivero, Arregui and Frąckowiak 2010) consists of the fact that manner modification, either in the form of a regular manner adverb, or included in the semantics of the lexical verb, must appear in such structures. Rivero, Arregui and Frąckowiak (2010) link this feature with the structural property of ISCs (see (9) below). Hence, we have (1) above with the lexically specified feature of inadvertence (in semelfactive, involuntary action), (2) with situationally specified circumstances (by accident), and (3) with an overt adverbial modification. Below we would like to additionally illustrate this requirement with a revealing pair of examples8 taken from Rivero, Arregui and Frąckowiak (2010: 711):

(6) *Zapaliło mi się sofę.

lit.up.NEU I.DAT REFL sofa.ACC

‘I set a sofa on fire by chance.’

(7) Zaprószyło mi się ogień w łóżku. I.DAT REFL fire.ACC in bed

‘I set a sofa on fire by chance.’

The acceptability judgments of these two sentences are in contrast, as in (6) the verb does not imply in itself the accidental manner of the action, and thus the action has to be adverbially specified for our ISC; as it is not, ungrammaticality results. (7) contains the verb which in itself specifies the accidental manner of the action and thus no adverbial modification must be placed in the structure for the sentence to be acceptable and fully grammatical.

Another feature of these ISCs, as discussed in Rivero, Arregui and Frąckowiak (2010) concerns the lack of control of the dative argument over the event described by the structure. In the abovementioned examples of ISCs, the situation has occurred due to external circumstances, and not because of the volition of the argument in the Dative case. In this light, another feature of ISC structures seems to be quite unexpected; namely, the dative argument seems unable to be linked to inanimate forces. This limitation is duly noted by Rivero, Arregui and Frąckowiak (2010: 711) but left unexplained:9

(8) *Słońcu się wzeszło.

sun.DAT REFL came out.NEU

‘The sun (somehow) came out.’

Consequently, the features to be accounted for in ISCs are: the occurrence of the dative argument in the absence of the nominative subject, the default verbal form showing no agreement, the necessity of manner modification, the human/animate character of the dative argument, which, however, does not exert any control over the event and, additionally, the presence of the resumptive się in such structures.

4. Rivero, Arregui and Frąckowiak’s (2010) analysis of ICSs in Polish

Rivero, Arregui and Frąckowiak (2010) claim to deal with ICS structures in Polish in the larger context of ‘out of control’ constructions, but as most of their analysis is devoted to, and based on, the Polish data, we feel that these data can be re-analysed on their own against the system of the Polish language to see how convincing their solutions are in this light.

The pivot of Rivero, Arregui and Frąckowiak’s (2010) analysis lies in the silent circumstancial modal they introduce, which heads an Applicative Phrase with an applied argument in the Dative filling the position of the specifier, and a Tense Phrase complement, which in turn contains the resumptive clitic się. Another obligatory element is the Manner Phrase. Other details will become clearer as we go along. Below we include the structure (p. 706) for the sentence: Jankowi tańczyło się dobrze ‘Janek. DAT danced.NEU REFL well’.




Manner Phrase

[[TP]] = λ10 x. λ e. λw. (danced CM TP dobrze λ x. λ e.λw.( good(x)(e)(w))

(e)(w) & Agent (x,e,w)) VoiceP


sięi VP

tańczyło . λ e. λw. (dance )(e)(w))11

The presence of the Applicative Phrase accounts for the introduction of the dative argument with a loosely specified semantic function, as it is a high applicative and such variation is precisely what Pylkkänen (2008) predicts for high applicatives. The differences in the function of the dative arguments in (1), (2) and (3) above, which we have discussed in Section 3, tally with such a structure. The fact that the phrase is headed by a modal accounts for the not-quite finite nature of the lexical verb, i.e. the fact that it is in the third person singular neuter, irrespective of the context. The presence of the manner phrase accounts for the necessity of manner modification in such structures, and also for the lack of control of the human element over the action as the manner is otherwise regulated. The requirement of the human interpretation for the dative argument is taken from Chierchia (1995) and especially Rivero and Sheppard (2003), where się is viewed as a variable, whose essential feature is human presupposition. As the reflexive pronoun, the external argument of the active voice (represented in the structure as i) and the applied argument are coindexed, the human factor is to be expected. The modal is human-dependent (because of the personal dative and the personal resumptive się). As a result, ISCs cannot appear e.g. with unaccusatives:

(10) *Żelazu topiło się łatwo

iron melted.NEU REFL easily

The modal has some more features, though they are only very briefly mentioned in the paper, while their more in-depth description may be found in Rivero (2009).12 The modal has universal force, as in must, it is a circumstantial modal, i.e. it is based on contextually identified facts, it is manner oriented as one of its arguments is the Manner Phrase, which, according to Rivero, Arregui and Frąckowiak (2010) means that the manner of the action has to be specified in ISCs either lexically, contextually or through a manner adverb:

(11) Marcie pisało się dobrze.

Marta.DAT wrote.NEU REFL well

‘Marta (by chance) wrote well.’

In (11) manner is specified by the adverb, in (1) above sneeze implies involuntary manner, in (2) the context prompts that something happened by accident. If no such manner adverb is suggested, ungrammaticality results:

(12) *Tańczyło się im

danced.NEU REFL them.DAT

Moreover, the structure in (9) implies that the dative argument has no control over the manner of the action13 and that the manner refers to the perception of the event by the applied argument and not to the event: Marcie pisało się dobrze (see (11) above) does not mean that she wrote good stuff, cf. Marta pisała dobrze ‘Marta.NOM wrote well’.

The analysis presented is concise, links the structure with the semantic properties very well, and describes the necessary data, explaining some phenomena that had evaded explication up to now, like the necessity of having the manner adverbial with such constructions, which was duly noted (see Rivero and Sheppard 2003), but not interpreted. Although we admire the elegance of the solution, we feel that some of its claims are not justified enough, especially if considered in the overall context of the Polish language. Rivero, Arregui and Frąckowiak (2010) take out this small portion of the data without heeding the broader context of the language structure.

5. Problems with Rivero, Arregui and Frąckowiak’s (2010) analysis

When considered in detail, Rivero, Arregui and Frąckowiak’s (2010) analysis presents many problems. First of all the notion of the circumstantial modal (CM), the pivot holding up the whole analysis, is an ad hoc element in the Polish system. For one thing it has no overt realisation and its presence is exclusively felt due to the influence that it exerts over the sentence interpretation and structure. It does not even constitute an affix (as it does in the data with which the authors contrast their analysis for Polish). Of course, the existence of various empty categories and empty morphemes is widely recognised in the literature,14 but such entities burden the linguist with a strong obligation to be extra careful and compile abundant argumentation supporting the existence of the said category. Polish does not show strong arguments for claiming that TPs of the type specified in (9) really possess a modal dominating the TP.

    1. Arguments against an empty modal in ISCs in Polish

In this subsection we will present some reasons why the existence of the modal element in Polish is doubtful.

5.1.1.Circumstantial and non-circumstantial readings of ISCs

First of all, we would like to draw the reader’s attention to the meaning of the phrases resembling ISCs, but just without the applied argument. Such phrases could be expected to possess the same structure as in (9) above, minus the applied argument, but with the modal over the TP (together with its manner modifier). Such sentences, though, as (13) below shows, do not necessarily have the modal meaning attributed to the phrase in (9):

(13) Tańczyło się dobrze

danced.NEU REFL well

The sentence in (13), unlike the one in (9), is ambiguous between two readings – one with the meaning characteristic of the presence of the circumstancial modal: the dancing was good because of independent reasons – and another, non-modal reading. The second reading allows human control over the now implicit argument: the quality of the action in the TP depends on the human element implied in the structure (with the help of the resumptive się). This meaning is at least equally viable as the circumstancial one:

(14) Tańczyło się dobrze, bo długo trenowaliśmy.

danced.NEU REFL well, as long practiced.1stPL Past

‘We danced well as we had been practising a long time.’

(14) is an equivalent of (15) below:

(15) Tańczyliśmy dobrze, bo długo trenowaliśmy.

dance1stPL PAST. well, as long practised.1stPL PAST

‘We danced well as we had been practising a long time.’

Consequently, as the presence of the modal in Rivero, Arregui and Frąckowiak’s (2010) model15 is essential for suspending the logical subject’s control over the circumstances, then, in the light of the interpretation in (14), the modal’s presence in (13) is at least spurious; consequently, its place in (9) also becomes doubtful.

5.1.2. True modal ISCs in Slovenian

The presence of the modal element as such in these structures has been contended for other reasons in Rivero and Sheppard (2003: 97). They supply two structures for ISCs, one for Polish and one for Slovenian, and then their modal force is examined and compared. They find out that the modal meaning exists in Slovenian, but not in Polish:

(16) Tę książkę czytało mi się z przyjemnością. (Pol)

this book. ACC read.NEU I.DAT REFL with pleasure

‘I read this book with pleasure.’


(17) Danes dopoldne se mi je jedlo jagode. (Slo)

today morning REFL I.DAT be.3S eat.NEU strawberries.ACC

‘This morning I felt like eating strawberries.’

This interpretation runs as follows (Rivero and Sheppard 2003: 97): ‘[16] denotes a past eventuality with the Dat as (involuntary) Agent. [17] is a modalized statement, not a past eventuality, and the Dat resembles an (involuntary) Experiencer. Thus, the speaker who truthfully utters the Pol sentence must have read a book, while the speaker who utters the Slo sentence need not have eaten strawberries.’ If we understand it correctly, the modality is relevant to the latter interpretation, and non-existent for Polish ISCs. And again, a similar thought is repeated later (Rivero and Sheppard 2003: 137), with respect to other data, which shows that Polish needs an overt modal to express a modal idea: ‘Pol [18] and Slo [19] each expresses the idea that John is sleepy, but the Pol version must be equipped with an overt modal and the Slo version need not be (…)


Jankowi chce się spać (Pol)

John.DAT want.3S REFL sleep.INF


Janezu se spi (Slo)

John.DAT REFL sleep.3S’

As we see from the above examples, in earlier versions of the same theory the very same data is denied the modal overtone, which is played up in the current version. We believe that this might undermine the soundness of the modal analysis of such structures as well.

5.1.3. ISCs in Polish do not share the formal features of other Polish modal-like structures

The examples above prompt one more problem with proposing the modal element in Polish ISCs. As we have seen in (18) above, in Polish the usual verbal form after a modal-like operator is the infinitive, and not the 3rd pers. sg. neut. In this group of operators we may include (see e.g. Grochowski 1984: 236–237) uninflected verbs like: można ‘can’, trzeba ‘must’, wolno ‘may’, and voluntative verbs such as: chcieć ‘want’, pragnąć ‘desire’, da się ‘it is possible’. In each of these cases it is the infinitival verbal form which follows the element with modal meaning. If these uninflected verbs and voluntative verbs are indeed elements semantically close to the proposed CM,16 then we may expect that their distribution would, at least in some cases, partially resemble that of CMs. Such a parallelism does not exist, as the examples in (20) below show:



Można dobrze tańczyć.

can well dance.INF

‘One/people can dance well.’


Da się to zjeść.

can it eat.INF

‘One/people can eat it.’

It has been pointed out to us17 that the form by (the conditional [irrealis] forming particle),18 which is the closest to an overt modal that Polish has, can appear with 3rd pers. sg. neut. verbal forms:

(21) Tej muzyki by się słuchało.

this music.GEN IRREALIS REFL listen to.3rd SG NEU PAST

‘This music could be listened to’.

However, here the appearance of the 3rd person default verb is very limited as the lexical verb has to appear in the past form, so its distribution is restricted, unlike in ISC constructions, which appear in all tenses. Compare the examples below:

(22) Tej muzyki *by się słucha

this music.GEN IRREALIS REFL słucha 3rd SG NEU PRESENT


(23) Tej muzyki nam się słucha dobrze.

this music.ACC we.DAT REFL listen to.3rd SG NEU PRESENT well

‘We can listen to this music with pleasure.’

If, however, we have the structure with both: the modal semantic predicate and the irrealis particle ((24) below), the lexical verb appears in the infinitive:

(24) Można by słuchać muzyki.

can IRREALIS listen to.INF music.ACC

‘This music can be listened to.’

Consequently, it does not have to be the modal element by that requires the appearance of the neuter default form in (21) above.

Of course, the facts that ISCs in Polish may not necessarily convey the modal meaning, have unprecedented phonologically empty modal morphemes in Polish, and appear with a type of TP not really to be expected, do not, each taken in isolation, constitute strong evidence against Rivero, Arregui and Frąckowiak’s (2010) analysis, but taken all together, they may set us thinking.

5.1.4. Manner modification is not modal-dependent

In this subsection we will show why the complex of the Applicative Phrase headed by a modal with a Manner Phrase argument is not very convincing or indispensable. If structures like Jankowi tańczyło się dobrze (see (9) above) contain both the applicative and the modal element, then we may assume that sentences without the applied argument, but otherwise identical, have the modal in their structure too, especially as some of the allegedly modal properties19 obtain for these sentences as well, e.g. human control, lack of these structures with inchoative verbs, etc.20 Yet, contrary to our expectations, applicative-less structures do not require adverbial modification:

(25) Tańczyło się. ‘Somebody danced.’

(26) Tańczyło się dobrze. ‘The dancing was good.’

(27) Tańczyło się do rana. ‘The dancing went on until morning.’

Unlike ISCs, the examples above show no sensitivity to adverbial modification, even though other properties attributed to the modal presence are preserved. Thus, we feel that, whatever causes the limitations described in Section 5.1.3, it certainly is not the CM.

    1. ISCs in Polish are not human dependent

The speculative nature of the presence of a modal element in the structure of Polish ISCs is not the only problem of Rivero, Arregui and Frąckowiak’s (2010) analysis; some of the properties that the authors attribute to the modal are (p.705) ‘hardwired’ into the whole structure of ISCs and therefore, if contended, make the whole derivation less convincing.

CMs are said to possess a human element in their interpretation.21 However, its presence is also doubtful. Consider a similar structure illustrated by (28) and (29) below. The structure possesses a lexical verb in a form identical to that in ISCs, and a dative argument signaling the Applicative Phrase, so it is possible that this structure shares some elements with ISCs.22

(28) Złamało mu nogę.

broke.NEU he.DAT leg.ACC

‘Something broke his leg.’

(29) Urwało mu rękę.

tore off.NEU he.DAT arm.ACC

‘Something tore off his arm.’

If the element deciding about the impersonal reading of the structures in (28) and (29) is the same as in ISCs – a modal element over the TP then it could be the very same modal as in ISCs (as the unnecessary multiplication of empty categories is quite costly). If so, then we could expect that the structure would have a human Agent in its reading, as the CM is pervaded by human characteristics. This is not the case. Totally the opposite, the Causer here is implied as some inanimate, non-human force. For instance, in the case of (28) the force could be a train, or a saw, but not a giant.

The constructions such as the ones above can be waived as they may not share the sensitive part of the structure with ISCs after all. Still, the supposedly human nature of CMs remains doubtful within the area of ISC constructions proper, where we find utterances like in (30):

(30) Słońcu wzeszło się dziś wcześnie.

sun.DAT rose.NEU PAST today early

‘The sun rose early today.’

They are grammatical, although slightly poetical. One may see in this example an anthropomorphic approach to the forces of nature, but, even so, such examples would suggest that the human reading of CMs is not structure dependent as Rivero, Arregui and Frąckowiak (2010) claim, but that it results from some pragmatic, easily overridden (weaker) constraints.

Even better, there are cases of non-human readings that cannot be reduced to the metaphorical treatment of some arguments. Consider the examples below:

(31) Przykszy mi się strasznie.

pine (after). NEU PRESENT I.DAT REFL awfully

‘I pine after something awfully.’

(32) Nudzi mi się okrutnie.

bore. NEU PRESENT I.DAT REFL awfully

‘I am bored to death.’

(33) Dłuży mi się niewiarygodnie.

tarry.NEU I.DAT REFL unbelievably

‘I am unbelievably bored.’

The examples in (31), (32) and (33) again suggest the existence of external non-personal circumstances exerting an influence on the personal Theme in the Dative: there is nothing personal about the modal element, if indeed such exists.

6. An alternative analysis of ISCs

Is the modal element really necessary in ISCs, is it unique to such structures, or can ISCs’ properties be explained in some other way? Such questions need to be asked and answered at this point. One of the facts that the CM purports to explain is the presence of manner modification in ISCs. Rivero, Arregui and Frąckowiak (2010) suggest that in the ISC structure the Manner Phrase is introduced by the modal as its argument. Consequently, it is an argument introduced higher than in a TP and thus it may appear in ISCs, even if a particular manner modifier is ungrammatical when it directly accompanies the lexical verb inside the TP. This high structural position accounts very well for the contrast in grammaticality between the utterances in (34) and (35) below (Rivero, Arregui and Frąckowiak (2010: 708):23

(34) Basia mieszka u siostry ‘Barbara lives at her sister’s’

(35) Basia*dobrze mieszka u siostry ‘Barbara *well lives at her sister’s’

(36) Basi dobrze się mieszka u siostry.

Barbara.DAT well REFL live.NEU PRESENT at sister.GEN

‘Barbara lives well at her sister’s.’

Again, we are not sure that such sentences argue in favour of their structure if we consider the system of the Polish language; the same effect can be noted in other sentences with applied arguments introduced by applicative heads as in (37) below, but these sentences give us absolutely no grounds for suspecting the presence of any modal element in them:24

(37) Basia dobrze sobie mieszka u siostry.

Barbara well applied argument.DAT live.3rd SG PRESENT at sister GEN

‘Barbara lives well at her sister’s.’

The sentence above contains an applicative construction with the dative argument sobie coindexed with the subject of the sentence. This points to the presence of an applicative head in the structure of (37), like in Pylkkänen (2008), or at least to some portion of the structure introducing such a dative (applied) argument. This, in turn, proves that the presence of the manner modification over the TP is not conditioned by any CM (even if we believe that there are CMs in ISCs), but by the presence of the applicative portion of the structure, whatever the structural details.

Accidentally, there is much more to be discovered about the link between the applicative phrase and the manner modification25 and we feel that it is rather the control properties that have a role to play in what is going on in the structures with the applied argument. Note in the examples (38) – (44) below that the applicatives in ISCs and sobie applicatives accept adverbial modifies differently, depending on the applicative type and the lexical verb:

(38) Mieszkam sobie dobrze u siostry. ‘I live well at my sister’s.’ (sobie applicative)


(39) Mieszka mi się dobrze u siostry. ‘I live well at my sister’s.’ (ISC applicative)

As (38) and (39) above show, both applicative structures are grammatical with dobrze as a modifying element. The situation gets complicated, though, if we take an action verb, like in the examples below:

(40) *Tańczyłam sobie dobrze. ‘I danced well.’ (sobie applicative)

(41) Tańczyłam sobie do rana. ‘I danced until morning.’ (sobie applicative)


(42) Tańczyło mi się dobrze. ‘I danced well.’ (ISC applicative)

(43) *Tańczyło mi się do rana. ‘I danced until morning.’ (ISC applicative)

(44) Tańczyło mi się dobrze do rana. ‘I danced well until morning.’(ISC applicative)

(44) above reveals that the ungrammaticality of (43) is due to the fact that do rana ‘until morning’ does not count as a manner modification; as soon as this element is added, the sentence gains grammaticality. However, the ungrammaticality of (40) remains unaccounted for; it cannot be due to the fact that such structures accept no broadly understood adverbial modification (cf. (38)). Note, however, that the types of verbs we have in (38) and (40) differ significantly. The one in (38) contains what may be believed to be an unaccusative26 verb, while the one in (40) an unergative. If, in the structure of (38), in its deep configuration, we have no external argument, then dobrze must be controlled by some other element in the structure – it refers to the event itself. On the other hand, (40) contains an unergative lexeme with the external argument introduced by the voice head.

Let us imagine that one of the lexical properties of dobrze27 is that it has to have scope over the whole event with its all participants, while do rana modifies only the lexical verb with its internal argument/arguments. Thus, in Tańczył dobrze ‘He danced well’ dobrze scopes over the whole sentence; in Tańczyło mi się dobrze (42) the applicative argument, especially if we dispense with the modal, could be directly introduced by the applicative head and dobrze has the scope over the whole event as well:


Appl P


Appl.head Manner Phrase (dobrze)



Voice P

sięi VP


On the other hand, *Tańczyłam sobie dobrze may have the following structure:28


Voice P




Appl. head TP Manner Phrase



Note that, in such a case dobrze does not scope over the whole event as the external argument is too high up in the structure, divided from the TP by the applied argument – an optional participant in the simple active voice structure of the event.29 We suggested before that dobrze has to scope over the whole event; since in (40), and (46) it fails to do that, ungrammaticality results.

This analysis also explains why Mieszkałam sobie dobrze u siostry ‘I lived well at my sister’s’ is grammatical; the surface external argument in this structure originates as the internal argument of the unaccusative verb, so dobrze may then scope over the whole event, before the internal argument is externalised.

These remarks must be treated as a loose introduction, and by no means a final solution to the problem, but we feel it is the right track to follow.30 On the whole, we feel that the area of the interaction of applicatives, ISCs and adverbial elements awaits further extensive work.

7. Conclusion

In this text we have considered the interface of semantic interpretations, syntactic structures as revealed by co-occurrence limitations and lexical representations. The material consists of Involuntary-state constructions and related applicatives in Polish. We have analysed these together with their adverbial modifications, which vary with respect to the lexical verb present in the structure. The difference between unaccusative and unergative verbs has been found to be significant for the analysed structures, as well as for the lexical properties of the adverbs like dobrze ‘well’.

We found a current analysis of ISCs, which proposes the structure with a circumstantial modal, to be unsatisfactory as it crashes when confronted with Polish material from related structures. Instead, we have presented an alternative account where interesting facts about ISCs and other similar constructions result from the applicative module present in strategic places of the relevant syntactic structures.


Bułat, E. 2004. Reflexive verbs as null object licensers in Polish. Similarities between the reflexive się and small pro. Poznań Studies in Contrastive Linguistics 39: 29–38.

Chierchia, G. 1995. The variability of impersonal subjects. In E. Bach, E. Jelinek, A. Kratzer, and B. Hall Partee (eds.), Quantification in natural language, 107–143. Dordrecht: Kluwer.

Chung S., and A. Timberlake. 1985. Tense, aspect, and mood. In T. Shopen (ed.), Language typology and syntactic description. Volume III. Grammatical categories and the lexicon, 202–258. Cambridge: Cambridge University Press.

Comrie, B. 1985. Causative verb formation and other verb-deriving morphology. In T. Shopen (ed.), Language typology and syntactic description. Volume III. Grammatical categories and the lexicon, 309–348. Cambridge: Cambridge University Press.

Grochowski, M. 1984. Składnia wyrażeń predykatywnych [Syntax of predicative expressions]. In Z. Topolińska (ed.), Gramatyka współczesnego języka polskiego. Składnia [Grammar of Modern Polish. Syntax], 213–299. Warszawa: PWN.

Heim, I., and A. Kratzer. 1998. Semantics in Generative Grammar. Malden, Mass: Blackwell Publishers Inc.

Dixon, R. M. W., and A. Y. Aikenvald. 2000. Changing valency. Case studies in transitivity. Cambridge: Cambridge University Press.

Kardela, H. 1985. A grammar of English and Polish reflexives. Lublin: Uniwersytet Marii Curie-Skłodowskiej.

Kratzer, A. 1996. Severing the external argument from its verb. In J. Rooryck, and L. Zarig (eds.), Phrase structure and the lexicon, 109–137. Dordrecht: Kluwer.

Kratzer, A. 1991. Modality. In A. von Stechow, and D. Wunderlich (eds.), Semantics: An international handbook of contemporary research, 639–650. New York: Walter de Gruyter.

Kratzer, A. 1981. The notional category of modality. In H.-J. Niekmayer, and H. Rieser (eds.), Words, worlds, and contexts, 38–74. Berlin: Walter de Gruyter.

Landau, I. 2010. The explicit syntax of implicit arguments. Linguistic Inquiry 41(3): 357–388.

Laskowski, R. 1984. Kategorie morfologiczne języka polskiego – charakterystyka funkcjonalna [Morphological categories of the Polish language – functional characteristics]. In R. Grzegorczykowa, R. Laskowski, and H. Wróbel (eds.), Morfologia [Morphology], 121–163. Warszawa: PWN.

Levin, B., and M. Rappaport Hovav. 1995. Unaccusativity. At the syntax-lexical semantics interface. Cambridge, Mass.: MIT Press.

Malicka-Kleparska, A. to appear. High applicatives in Polish. In E. Cyran, H. Kardela, and B. Szymanek (eds.), Volume in Memory of Ed Gussmann, .

Marantz, A. 1984. On the nature of grammatical relations. Cambridge, Mass.: MIT Press.

Pylkkänen, L. 2008. Introducing arguments. Linguistic Inquiry Monograph 48. Cambridge, Mass.: MIT Press.

Perlmutter, D. 1978. Impersonal passives and the unaccusative hypothesis. In Proceedings of the Fourth Annual Meeting of the Berkeley Linguistic Society, 157–189. Berkeley: Berkeley Linguistic Society, University of California.

Rivero, M. L. 2009. Intensionality, high applicatives, and aspect: involuntary state constructions in Bulgarian and Slovenian. Natural Language and Linguistic Theory 27: 151–196.

Rivero M. L., A. Arregui, and E. Frąckowiak. 2010. Variation in circumstantial modality: Polish versus St’át’imcets. Linguistic Inquiry 41(4): 704–714.

Rivero, M. L., and M. Milojević Sheppard. 2003. Indefinite reflexive clitics in Slavic: Polish and Slovenian. Natural Language and Linguistic Theory 21: 89–155.

Spencer, A. 1991. Morphological theory. An introduction to word structure in Generative Grammar. Cambridge, Mass.: Basil Blackwell Inc.

Szober, S. 1962. Gramatyka języka polskiego [A grammar of

the Polish language]. Warszawa: Państwowe Wydawnictwo Naukowe.

Szymanek, B. 2010. A panorama of Polish word-formation. Lublin: Wydawnictwo KUL.

Wróbel, H. 1984. Słowotwórstwo czasowników [Word-formation of verbs]. In R. Grzegorczykowa, R. Laskowski, and H. Wróbel (eds.), Morfologia [Morphology], 467–511. Warszawa: PWN.

Zwicky, A. 1985. Clitics and particles. Language 61: 283–305.

1 We will also take theoretical notions liberally from Rivero (2009) and Rivero, Arregui and Frąckowiak (2010).

2 Anticausative derivations reduce the number of arguments appearing with a verb, disposing of the syntactically stated direct object. See e.g. Comrie (1985: 319–322), Dixon (2000: 6–11).

3 Się does not have separate word stress and it must accompany a verbal element, but unlike true affixes it can precede or follow the verb more or less freely, e.g. On się pakuje ‘He is packing up’ vs. On pakuje się od dwu godzin ‘He has been packing up for two hours’. For details see, e.g. Szober (1962), Wróbel (1984), Kardela (1985), Bułat (2004), Szymanek (2010) etc.

4 If taken rigorously, the principle should imply that there are no lexically empty terminal nodes, the result deserving careful consideration.

5 Although they may also function as regular 3rd person singular neuter agreement verb-forms, e.g. Dziecko poszło do lasu ‘The child went.3rdSG NEU PAST into the woods’.

6 See e.g. Laskowski (1984: 146–148).

7 Rivero, Arregui and Frąckowiak (2010) do not consider such structures, as they are not involuntary states.

8 We have different grammaticality judgments concerning the example in (6) than their authors, who mark both of them as grammatical, while we find the first of them odd. However, with the example in (6) being ungrammatical, the point is illustrated even more clearly.

9 Notice, however, that in the example in (8) Rivero, Arregui and Frąckowiak (2010) do not supply any manner adverb. The sentence: Słońcu wzeszło się dziś wcześnie ‘The sun rose today early’ sounds much better and according to our judgment would be grammatical, if somewhat poetical. It is true, however, that Polish grammarians claim that such structures imply personal performer of the action, cf. e.g. Laskowski (1984: 147 ): ‘(…) konstrukcje nieosobowe na się i m p l i k u j ą i s t n i e n i e osobowego wykonawcy czynności lub nosiciela stanu czy procesu (…)’ [impersonal constructions with się imply the existence of a person performing the activity or a bearer of a state or process tr. mine A.M.-K.]. This remark appears in the context of dativeless structures though.

10 Exact semantic representations cannot be introduced within the limits of this paper in any great detail. We will discuss the ones that are immediately necessary. For more information refer to Heim and Kratzer (1998: 34–39). To put it very crudely, lambda expressions stand for functions with arguments specified as variables (e.g. x, y, z) and values specified for these variables.

11 Rivero, Arregui and Frąckowiak (2010: 705) sum up the structure in the following way: ‘(…) we find a meaning constructed with resources outside the inflectional system and verbal paradigm, namely, with a dative in a new type of high applicative with modal properties that stands above TP and thus dominates the inflectional space of the clause.’

12 Rivero (2009) models her analysis of modal elements on Kratzer (1981, 1991).

13 Rivero and Sheppard (2003) and Rivero (2009) link the property with the fact that the argument is in the dative and it is not in TP as such, so consequently, although it mostly plays the part of a logical subject, some of its properties, for instance control, are suspended.

14 See e.g. Landau (2010) for a recent and convincing proposal, with an extensive inventory of such categories.

15 Rivero, Arregui and Frąckowiak (2010: 709) write: ‘(…) the ISC in [Jankowi tańczyło się dobrze] is true in a world iff the worlds that satisfy the modal base (i.e. those matching the actual world with respect to relevant circumstances), including the presupposed event of Janek dancing, are also worlds in which the event was pleasurable for Janek.’

16 CM stands for circumstantial modal.

17 Prof. Anna Bondaruk [personal communication] has supplied us with another example of 3rd pers. sg. neut. used with by: Aby żyło się lepiej ‘Let there be a better life for us’, which is the equivalent of a subjunctive expression in English, and to which the following limitations also apply. These phenomena certainly await some additional investigations.

18 See Zwicky (1985), Spencer (1991), Chung and Timberlake (1985).

19 Of course, the problems with such assumptions discussed in connection with (13) and (14) above and Section 5.2 below will be relevant to these examples as well.

20 *Topniało się dobrze ‘Something melted well’ is as ungrammatical as *Topniało się dobrze metalowi ‘The metal melted well’ with an applied argument.

21 Rivero, Arregui and Frąckowiak (2010: 705) claim that ‘(…) the ISC modal must be human dependent/personal, a requirement that we derive from its syntax. The ISC modal heads an Applicative Phrase with a dative specifier and a clausal complement: a Tense Phrase (TP). The TP contains a reflexive human pronoun resuming the dative: the so-called impersonal się. ISCs, then, display sensitivity to the subject based in dative and reflexive marking hardwired into the interpretation of the modal, which must be personal’.

22 As we have shown in section 5.1.1, the modal meaning is not necessary with Rivero, Arregui and Frąckowiak’s (2010) CMs.

23 Note, however, that Rivero, Arregui and Frąckowiak (2010: 708) support their analysis by giving an example of a structure with two adverbial elements, one allegedly high in structure, in the modal-dependent position, the other low, in the TP: Dobrze Jankowi tańczyło się fatalnie ‘It felt good to John that he danced awfully’. The sentence is outright ungrammatical in Polish though, so it cannot support their analysis. It does not disprove it either though, as there may be some other limitations involved rendering it ungrammatical than the lack of adverbial modification higher than in the TP. We believe that the ungrammaticality of *Dobrze Jankowi tańczyło się fatalnie is connected with the fact that both layers of the structure are human oriented in the sense of Rivero, Arregui and Frąckowiak (2010), and that it is the same human in both cases, so a logical clash results in the interpretation.

24 For a recent analysis of such applicatives see Malicka-Kleparska (to appear).

25 In contrast to the doubtful relationship of a CM and a manner phrase.

26 The distinction between unaccusative and unergative verbs is very well motivated since the times of Perlmutter’s (1978) Unaccusative Hypothesis. For the purposes of this study we adopt the understanding of unaccusative and unergative terms taken from Levin and Rappaport Hovav (1995: 3) as D-structure configurations: Unergative verb: NP[VPV], Unaccusative verb___[vpV NP/CP].

27 Rivero, Arregui and Frąckowiak (2010: 708) allow for the possibility that ‘(…) there may be more than one manner in ISCs. (…) dobrze ‘well’ serves as the argument of CM, while fatalnie ‘terribly’ operates within the TP and describes the dancing.’ We would like to claim that do rana, although it is not manner modification, operates within the TP.

28 The general idea of the applicative structure with an abstract applicative head is taken from Pylkkänen (2008); the modifications result from our considerations concerning ISCs and the related structures.

29 As predicted *Tańczyłam im dobrze ‘I danced well for them’ does not sound any better.

30 The presence of similar problems has been signalled in Rivero and Sheppard (2003), but no solution was offered.

© 2019
wyślij wiadomość

    Strona główna