Theoretical implications of the prefixation of Polish change of state verbs

1 Introduction

This article is written in the framework of Distributed Morphology (DM), in which the only generative component is morpho-syntax (Alexiadou and Anagnostopoulou 2004, Alexiadou 2010, Alexiadou et al. 2015, Borer 2005a, b). This means that the major properties of language structures are coded in constructions revealing layered organization, with particular layers of structure representing different projections contributing distinct elements of grammatical functions. The constructions are filled with morphological material; then, the whole structure is interpreted by the semantic component of language. A corollary of this view on grammar is that the whole basis of semantically relevant information has to be contributed by the shape and layered complexity of the structure itself, aided by the idiosyncratic information introduced by particular morphemes. Such a view of grammar puts significant stress on the (re)construction of relevant and general language structures as structures account for particular regular or sub-regular language phenomena.

In the present text, we concentrate on two seemingly unconnected regularities in Polish, viewed here as interconnected: the dual telic and atelic nature of synthetic change of state (COS) verbs and the resistance of the said verbs to take overt lexical prefixes. In our analysis, we will argue that these regularities are connected in a direct way, i.e. the structures deriving synthetic COS verbs disallow the insertion of overt lexical prefixes. As regularities in the system of Polish, the phenomena will be presented in terms of structural coding.

The first phenomenon on which we want to concentrate is a particular sub-regularity in the distribution of lexical prefixes (Section 2.1) in Polish with the class of COS verbs. While these verbs can appear with perfectivizing prefixes representing pure perfectivizer (1a) and superlexical (1b) subclasses, the literature of the subject largely overlooks the lack of lexical prefixes in the structure of the relevant verbs ((1c) Jabłońska 2004).

Apart from this compositional (ir)regularity, synthetic COS verbs manifest another interesting feature: Predications including unprefixed COS verbs show both telic and atelic properties. Both adverbial temporal modifiers ‘in X time’ (2a) and ‘for X time’ (2b), the first appearing in telic predications, the other – in atelic ones (Verkuyl 1972, Dowty 1979, Dik 1997, Borik 2002) – can accompany unprefixed COS verbs:

In (2b), we supply two temporal adverbial expressions occurring in atelic predications because the use of the prepositional phrase, directly mimicking the phrase proposed for English, might be problematic: Willim (2006, 176) in ftn. (2) claims that the adverbial introduced with przez ‘for’ may have both simple duration and time span readings in Polish. To eliminate time span readings, bare adverbials can be used. However, in many contexts, bare adverbials sound colloquial, though not ungrammatical. An additional problem may arise when bare adverbials appear in sentences with transitive verbs, when the verbs are used in an absolute phrase, i.e. they do not have overt objects. In such cases, bare adverbials may be misinterpreted as direct objects of a verb (see 14a). This factor may contribute to native speaker’s preference for prepositional adverbial phrases.

The adverbials coding short stretches of time that can be expressed with frequently used, one-word expressions, fare much better as bare adverbials. Lexical items coding long stretches of time, like stulecie ‘century’, tysiąclecie ‘millennium’, cannot be used as bare adverbials at all. Whenever possible, we will be using bare adverbials to test the telicity of predications.

The telic/atelic dual nature of COS verbs in (2) requires an explanation within the Generative framework of DM as it has been under-researched or not researched at all. Both prefixation facts and telicity phenomena will be targeted together in our article because they will be presented as interconnected.

Since our text deals with language facts involving syntagmatic phenomena concerning formal exponents at the lexical level of morpho-syntactic structures (i.e. the distribution of verbal aspectual prefixes), paradigmatic issues concerning verbal typology (synthetic COS verbs as a grammatically relevant class) and syntagmatic issues of super-lexical structures, which are bases of the semantic interpretation of a predication (perfective/imperfective, telic/atelic distinctions), in the first three sections, we will introduce the pivotal phenomena. Thus, Section 2 characterizes synthetic COS verbs in Polish. Section 3 adumbrates basic facts related to properties of Slavic aspectual prefixes, with particular stress put on Polish prefixes of this type. Section 4 provides a sketch of issues relating to telicity/atelicity and perfective/imperfective aspects in Slavic and Polish in particular. Section 5 presents the major issue of the gap in the distribution of lexical prefixes with COS verbs and presents problems arising from an existing analysis of the data (Jabłońska 2004). In Section 6, we propose an alternative analysis which links the limited distribution of aspectual prefixes with synthetic COS verbs to structural properties of homonymous telic and atelic verbal pairs. Section 7 introduces some problems parallel to the one pivotal for this text and briefly mentions an existing Nanosyntactic analysis of some COS verbs. Section 8 sums up the findings, pointing to more general theoretical implications of the model.

2 Synthetic COS verbs

The class of verbs that we will concentrate on in this text has been given various names in linguistic literature. Perlmutter (1978), Burzio (1986) and Levin and Hovav (1995) call them unaccusatives; Lakoff (1968, 1970), Parsons (1990) and Haspelmath (1993) refer to them as inchoatives – stressing the initiation of the COS that they code. Then, these verbs are also called decausatives by Fehrmann et al. (2011) – which underscores their relation with cognate causative verbs, and so do the terms ergatives (intransitive members of ergative pairs – Jones and Levine 2010) and anticausatives (Koontz-Garboden 2009). In more recent accounts, COS verbs are referred to as inchoative unaccusatives (Zdziebko 2019) and degree achievements (Taraldsen Medová and Wiland 2019). We will refer to them as COS verbs (following Levin and Hovav 2013) as this term seems the most neutral, stressing no necessary relationship with a transitive/causative counterpart and making no claims as to the initiation of the COS as a prerequisite to be counted as a member of the class. We define COS verbs as coding events with a single participant whose characteristics change in the event.

We will concentrate on synthetic COS verbs in Polish, i.e. such predicates whose whole morphological structure is realized by word-internal morphology, i.e. within the limits of a single word, whose individual morphemes cannot be re-arranged.^[2] This class is illustrated with examples in (3):

Examples in (3a) include verbs consisting of roots followed by verbal stem-forming morphemes quoted as -e-/-ej- (biel-e-ć ‘whiten-INF’ vs biel-ej-e-3SG.PRS) in sources devoted to Polish grammar (Laskowski 1984b, 189, Rubach 1984, Zdziebko 2019, Bloch-Trojnar 2023). The verbs terminate in inflectional endings. (3b) contains examples of COS verbs with stem forming -ną- plus inflection (Taraldsen Medová and Wiland 2019, Zdziebko 2019), while in (3c) roots of foreign origin take stem-forming -owa- and inflectional endings (Zdziebko 2019, Bloch-Trojnar 2023).

In spite of their morphological diversity, synthetic COS verbs behave in a uniform fashion while accommodating prefixes: they all resist the attachment of lexical prefixes.

In the next section, we will adumbrate some basic facts about classes of Slavic aspectual prefixes attached to verbs in order to build the theoretical background for our analysis.

3 Types of aspectual prefixes in Slavic

The types of aspectual prefixes in Slavic languages have been extensively discussed in linguistic literature, especially since the publication of a series of seminal works by Svenonius (2004a, b), Romanova (2004), Jabłońska (2004, 2007), Slabakova (2005) and Łazorczyk (2010). Their findings, in reference to the Polish material, will be summarized below.

Slavic aspectual prefixes can be divided into three classes: lexical, superlexical (Svenonius 2004a, b, Romanova 2004) and pure perfectivizers (Jabłońska 2004). Lexical (inner, low) prefixes are merged very early in the verbal structure, and they will be of special interest to this article since they are not encountered in COS verbs. Superlexical prefixes (outer, high) are merged relatively late in verbs, while pure perfectivizers occupy an intermediate structural position between lexical and superlexical prefixes. Superlexical prefixes and pure perfectivizers appear in COS verbs. One of the aims of this article is to explain in structural terms this distributional discrepancy. Below we will describe these classes of prefixes in some detail, including comments on their grammaticality with COS verbs in Polish.

4 Telicity/atelicity vs perfective/imperfective verbs

Since the phenomena connected with telicity or atelicity of predications and perfective and imperfective nature of verbs constitute regularities within the system of Slavic languages, they should find their reflex and explication in morpho-syntactic structures proposed within the system of DM. Although this article does not aspire to do justice even to major complexities connected with these phenomena, nevertheless a short introduction of the intricate and interesting relationships between the two grammar-pervading notions seems to be in order. A comprehensive outlook on the development in this area of study is presented in van Hout et al. (2005). Outside the framework of DM, this problem has been extensively studied in linguistic literature by Filip (2005, 2008, 2017), Fleischhauer and Czardybon (2016), and Kagan (2013), to name just a few more recent accounts.

Slavic languages, and Polish among them, illustrate all possible combinations of telic/atelic and perfective/imperfective propositions. To start with simplex verbs, i.e. the verbs that have no obvious derivational bases, it has been observed that while most of such verbs are imperfective and appear in atelic predications (13a), there are a few which are both perfective and telic. While telicity can be diagnosed by means of ‘in X time’/‘for X time’ test, perfective/imperfective forms of verbs are distinguished by a number of tests (see Schoorlemmer 1995, Filip 2000, Borik 2002, Romanova 2004). In Polish (Fleischauer and Czardybon 2016, 184–5) perfective verbs cannot complement phase verbs, such as skończyć ‘finish’, przestać ‘stop’ (13b), nor can they appear in the analytical future tense formed with the auxiliary być ‘be’ (13c). Just a handful of Polish simplex verbs appear in telic predications (13a) and are perfective (13b), e.g.: lec ‘lie down’, rzec ‘proclaim’,^[11] wziąć ‘take’ (Isačenko 1962 for Russian, Filip 2008 for Czech examples). The majority of simplex verbs show atelic properties (14a) and are imperfective (14b, c), e.g. pić ‘drink’, jeść ‘eat’, kraść ‘steal’. None of the above verbs can appear in the unchanged form in both telic and atelic propositions; i.e. they do not show the behaviour characteristic of COS verbs.

Morphologically more complex verbs show even more intricate patterns. Verbs prefixed with lexical perfectivizing prefixes (or pure perfectivizers) are perfective in a telic predication (15), while when these verbs are equipped with suffixes forming secondary imperfectives, they are imperfective (16a), but telic (16b).

At the same time, some prefixes, like e.g. superlexical po-, may not result in telicity of a predication (17a), although the verb is perfective (17b, c).^[12]

Similarly, prefixes expressing modal meanings (Willim 2006, 179) have been shown to derive perfective verbs which head atelic clauses, e.g. u-siedzieć ‘manage to sit still’, wy-żyć ‘manage to get by’.

Both language phenomena discussed here may influence also the clause structure in Slavic. Leaving aside the presence of distinct temporal modifiers and (im)possibility of their complementing phase verbs, we may observe that some unprefixed verbs may have absolute uses or appear with a nominal complement (19a), while their prefixed variants require the presence of the overt object (19b).

Similarly, telicity may pervade the structure of predication beyond the area of adverbial modification: It may influence the interpretation of the direct argument as definite (Czardybon and Fleischhauer 2014), which in some Slavic languages may be coded by the definite article (e.g. Bulgarian, Upper Silesian Polish).

The views on the interrelations between telicity and perfective marking in Slavic differ significantly in the literature on the subject. Filip (2003, 2008) stresses the fact that telicity correlates with perfectivity of the verbal stem as a whole and states that viewing telicity as having its source in the functional structure spelled solely by prefixes is misguided. Fleischhauer and Gabrovska (2019) link telicity with particular semantic make-up of prefixes, rather than with grammatical perfectivity as such. Fleishhauer and Czardybon (2016) observe that not all verbal prefixation results in telicity of a clause, while Filip (2008), as well as Czardybon and Fleischhauer (2014), claim that atelicity or telicity of clauses with imperfective verbs may result from larger (sentential) context. None of the sources, however, postulates the complete separation of the two phenomena in all areas of Slavic grammar. In this text, we will show that at least in the case of COS verbs in Polish perfectivity and telicity seem to be related, although not equalled, with consequences for the distribution of perfectivizing prefixes.

As this short introduction into the area of the interaction between perfective/imperfective verbs and telicity/atelicity of predications suggests, the problem is very complex and controversial. In this article, we will limit ourselves to these interactions and their structural representation which are relevant for COS verbs exclusively, as no theoretically relevant and at the same time true to data analysis of the whole verbal system of Polish can be undertaken within the limits of an article.

In the following sections, we will discuss Jabłońska’s (2004) attempt at explaining the lack of lexical prefixes with COS verbs, pointing to its shortcomings. This will be followed by our own rendering of the distributional facts as linked with the dual telic/atelic nature of predications with COS verbs.

5 Jablońska’s (2004) analysis

Although the regularity holding between COS verbs and lexical prefixes in Polish has been largely overlooked in DM accounts, one analysis of this area of grammar should be cited: Jabłońska (2004) observes that COS verbs in Polish do not appear with lexical prefixes.^[13] She also states that COS verbs can take pure perfectivizers and (rarely) superlexical prefixes. Her explanation of the limitation on the attachment of lexical prefixes relies on the claim that COS verbs are built on adjectives and adjectives do not take result phrases (RPs) as complements. Since lexical prefixes may contribute to the resultative meaning, she assumes (after Svenonius 2004a, b) that such prefixes are heads of RPs in the structure of morphologically complex verbs. Consequently, with the adjectival base within a complex verbal form, there is no place for the resultative projection, nor for the resultative head – occupied by a lexical prefix.

The presence of the RP in the morpho-syntax of telic verbs has been postulated by Svenonius (2004b, 206), who proposes the following structure for Slavic lexically prefixed verbs:

In his model, prefixes are introduced as heads of the RP. Since adjectives, heads of Adjectival Phrases, cannot be complemented with RPs (unlike verbs), lexical prefixes do not have an appropriate point of attachment and complex de-adjectival COS verbs never take them.

Jabłońska’s (2004) explanation has several drawbacks. First of all, it is doubtful whether COS verbs really contain Adjectival Phrases in their structures. Many COS -e-/-ej- verbs do not show any morphological relationship with adjectives: Instead, they have corresponding simple cognate nouns (21).

Some COS verbs do not have any related cognate simplex forms at all (21), so there is no morphology-based reason to treat them as de-adjectival either.

If the verbs in (21) and (22) possess any cognate forms functioning as NP modifiers (i.e. adjectival), these are passive adjectives showing morphological affinity to unaccusative verbs and formed with the morpheme -ł-, e.g.: z-babia-ł-y ‘like a woman’, z-ropia-ł-y ‘suppurated’, s-truchla-ł-y ‘deadened’, etc. (Cetnarowska 2000, 2002). A passive adjective consists of a verbal prefix, a stem, and the passive morphology marker (-ł-). Consequently, one may claim that passive adjectives are de-verbal, but not that COS verbs are de-adjectival. The explanation that the verbs in (21, 22) have adjectives in their derivational history, which precludes them from taking RPs as complements, simply does not work for such verbs, yet their behaviour with respect to lexical prefixes is precisely the same as that of COS verbs with corresponding simple adjectives.

The same situation exists for -ną- COS verbs: Many of them have no corresponding simple adjectives (23).

Again, some passive adjectival forms arise, e.g.: z-więdły ‘wilted’, wy-stygły ‘cold’, but these passive adjectives have internal morphologically complexity of unaccusative verbs, just like in the case of -e-/-ej- verbs in (22).

Then, COS verbs containing stem-forming (-iz)-owa- (3c above) tell the same story: They are built on non-native roots, which do not appear in simplex adjectives in Polish either. Out of this small group of COS verbs, only krystal-iz-owa-ć ‘crystallize’ has a corresponding simplex form, and this form is nominal: kryształ ‘crystal’.

All synthetic COS verbs, whether they have corresponding simpler cognate adjectives or not, behave alike with respect to lexical prefixation: they deselect lexical prefixes.

The only possible conclusion is that no Adjectival Projection is responsible for their behaviour, contrary to Jabłońska’s (2004) claim.

The above argument based on the absence of simplex adjectival cognates of COS verbs could be disregarded if we propose that COS verbs always go through an adjectival phase in their derivation, irrespective of existing cognates. However, no morphological support for such a claim is forthcoming as no morpheme characteristic of adjectives ever turn up in the structure of the verbs analysed above.

Another argument against a de-adjectival analysis of the gap in the distribution of lexical prefixes comes from the area of overt resultative complementation of COS verbs. Overt resultative phrases can occur in clauses headed by COS predicates, although they are not frequent, which is only to be expected if resultative meanings are in fact regularly expressed by lexical prefixation in Slavic (Svenonius 2004a, b). Svenonius (2004a, b) supports his analysis of Slavic lexical prefixes as heads of RPs drawing a parallel between such prefixes and various particles and prepositions in Germanic languages. He observes that particles and prepositions in Germanic languages, as well as lexical prefixes and prepositions in Slavic languages, are frequently homophonous and realize similar meanings. Thus, they may occupy similar positions in morpho-syntactic structures. Various resultative relations in Slavic can be expressed with the use of prepositional phrases, but more frequently they are lexicalized in the very prefixed verbs in the form of RPs headed by lexical prefixes.

Levin and Hovav (1995, 34) define resultative phrases in the following way:

“A resultative phrase is an XP that denotes the state achieved by the referent of the NP it is predicated of as a result of the action denoted by the verb in the resultative construction.”

Resultative phrases can appear with COS verbs (unaccusatives in Levin and Hovav 1995 terminology). Levin and Hovav (1995, 39) supply the following example:

Resultative expressions can also be found in Polish (Gulgowski 2013) with COS verbs (25).

As the examples in (25) show, both prefixed (25b) and unprefixed (25a) COS verbs in Polish can co-occur with phrases expressing results. Consequently, there is a space in the morpho-syntactic representation of COS events for RPs and the semantics of such events does not explain why COS verbs do not incorporate lexical prefixes as heads of such phrases. It follows that the line of reasoning which links the lack of lexical prefixation of COS verbs with ungrammaticality of resultative complementation of COS verbs cannot be upheld.

Jabłońska’s (2004) explanation concerns only the cases in which lexical prefixes have resultative meanings. It offers no insight into spatial relations expressed by them. Spatial modifiers appear regularly in COS clauses as the example in (26) shows, similarly spatial modifiers can appear with adjectives (27):

As the data in (26) and (27) show, the mere semantics of COS clauses does not preclude lexical prefixes realizing spatial relations from modifying such verbs, even if the verbs were de-adjectival.

We have shown above that the ungrammaticality of COS verbs with lexical prefixes in Polish cannot be explained as a result of their de-adjectival provenience, or, more generally, their inability to appear with resultative of spatial expressions. If lexical prefixes share some affinity to resultative and spatial expressions, as argued by Svenonius (2004a), then a different explanation for this gap in their distribution has to be proposed. Below we will put forward a structure-based explanation of this gap.

6 Structure-based explanation of the gap in the prefixation of COS verbs

In this section, we will consider and discard the possibility that the dual nature of COS verbs is due to homogeneity obtaining between open-class or closed-class adjectival bases of atelic and telic COS verbs (Section 6.1). Then, we will consider how contextual conditioning fares as the factor influencing telicity/atelicity of COS predications (Section 6.2). Finally, we will argue that distinct structures for telic and atelic COS verbs proposed within DM account for their morpho-syntactic environment, semantics, and the distribution of prefixes in this class of verbs (Section 6.3)

6.3 Structures for synthetic COS verbs

Since we have discarded solutions attributing the dual nature of synthetic COS verbs in Polish to contextual considerations, we would like to put forward a different claim: all synthetic COS verbs in Polish are in fact pairs of homonymous^[18] verbs: telic and atelic in each pair.

Telic verbs have the structure proposed for telic predicates by Svenonius (2004b, 206), with the proviso that the head of the RP is the zero morpheme. The presence of this morpheme is conditioned by the verbalizing morphemes -e-/-ej-, -ną-, -owa- (see 3), i.e. the morphemes which build verbal stems of synthetic COS lexemes. Consequently, other telicizing morphemes (lexical prefixes), which, in principle, could occupy the same structural position, are excluded. The selection of the zero morpheme precludes the attachment of other lexical telicizing morphemes to telic COS verbs since only one RP is available in the sub-lexical verbal structure.^[19]

(32) represents a schematic representation of telic COS verbs. The zero prefix is inserted in the position of the head of the RP. DP (determiner phrase) contains the argument in the RP (which surfaces as the subject of an unaccusative verb).

(32)

The presence of the RP makes the predication telic (33a), but it does not make the verb perfective, which is what the test for perfectivity shows (33b).

(33)
	a)	Róża		więdnie	w	godzinę.
		roses-NOM		wilt-IPFV	in	hour.ACC
		‘A rose wilts in an hour’.
	b)	Róża		będzie	więdnąć.
		Roses-NOM		will.be	wilt-IPFV.INF
		‘A rose will be wilting’.

Unlike overt lexical prefixes, the zero prefix does not carry the feature of perfectivity. However, the insertion of the zero prefix blocks the possibility of the insertion of overt lexical prefixes equipped with the feature [+perfective]. Consequently, zero-prefixed verbs are not perfective (33b). The insertion of the zero prefix explains why COS verbs take no overt lexical prefixes: There is just a single RP in the structure of a verb (32), and the zero prefix occupies its single head position. As overt lexical prefixes are inserted in verbal structures as heads of RPs and they cannot stack, they are excluded from COS verbs.

COS verbs prefixed with pure perfectivizers or superlexical prefixes are exclusively telic (34a) and perfective (34b), the feature [+ perfective] being contributed by superlexical prefixes or pure perfectivizers.

(34)
	a)	Róże	po-/z-więdły	w	godzinę/	*godzinę.
		roses-NOM	S/PP-wilt	in	hour-ACC/	hour
		‘Roses wilted in an hour’.
	b)	Róże	będą	*po-/*z-więdnąć.
		roses-NOM	will.be	S/PP-wilt-INF

Atelic COS verbs are simple structure-wise and consequently meaning-wise: They include neither the RP, nor higher aspectual projections. Thus, atelic COS verbs are never prefixed. The simple structure accounts for the simple semantics of such verbs. They refer just to the ongoing COS of the subject argument, without the culmination of the change. The structure for atelic COS verbs is provided in (35).

(35)

The above analysis of COS verbs within DM allows us to explain in terms of structural distinction between atelic and telic structures the distributional gap left by the absence of lexical prefixes with these verbs. The absence of lexical prefixes, but the presence of superlexical prefixes and pure perfectivizers is linked here with the shape of telic structures with the RP and atelic structures, without such a phrase in which these verbs can appear. The structure also allows us to represent the fact that atelic COS verbs, unlike telic COS verbs, can never be prefixed (which is not necessarily true about the verbal system of Polish as such – 17a and 18a).

This analysis of telic/atelic COS homonyms may have consequences for the treatment of telicity/atelicity and perfectivity/imperfectivity in Polish (and potentially in other Slavic languages) as separate phenomena in terms of structure. Telicity relies on the presence of the RP in a verbal structure, perfectivity/imperfectivity, on the other hand – on the presence of overt markers of various kinds (over lexical and superlexical prefixes, pure perfectivizers, secondary imperfectivizers). This is not to say that there are no interrelations between the phenomena. Telic structures may or may not include perfective verbs, the latter combination of properties being realized by zero-prefixed COS verbs devoid of superlexical prefixes and pure perfectivizers, but overt marking of verbal stems with lexical prefixes (except the stems including secondary imperfectivizers) correlates with telicity (Section 2.1).^[20] However, there can be no equation mark between the two concepts.

7 Excursus

Although we limit ourselves to analysing and representing regularities relevant to the verbal system of Polish, one may ask the question of whether similar regularities can be observed in other Slavic languages. It should be noted that we have observed the same regularity concerning the distribution of lexical prefixes with COS verbs in Russian. The class of COS verbs without the reflexive-like morpheme -sja (i.e. the verbs corresponding to Polish synthetic COS verbs) seems to co-occur with pure perfectivizers, e.g.: iz-goret ‘burn’, is-tlet ‘rot’, so-staret ‘get old’, u-potet ‘sweat’, etc. as well as with superlexical prefixes, e.g.: za-belet ‘begin to whiten’, ot-syret ‘begin to get wet’, po-blednet ‘begin to look white’, voz-goret ‘begin to be on fire’, ot-kipet ‘cease’, do-goret ‘burn out’. However, like in Polish, Russian COS verbs seem not to accept lexical prefixes, e.g. belet ‘whiten’ does not yield ‘nad-belet’, ‘podo-belet’, or ‘pre-belet’; i.e. there are no COS verbs derived with lexical prefixes specifying spatial orientation. Nevertheless, the issues relating to telicity/atelicity of the relevant verbs in Russian and their cognates in other Slavic languages, as well as generalization concerning the distribution of lexical prefixes with COS verbs would require a much more thorough study of such phenomena than we are prepared to attempt in this article.

Another issue tangential to the topic of this article concerns other classes of verbs which might show similar regular gaps in the distribution of prefixes. One remarkable example in Polish consists of verbs coding states discernible by means of senses (37).

Some of these verbs appear with the inceptive superlexical prefix za-, za-pachnieć ‘begin to exude scent’, za-śmierdzieć ‘begin to exude odor’ but no other prefixes are admissible. This gap in prefixation may not be attributed just to the fact that these verbs are states since other states admit a variety of prefixes, e.g.: spać ‘sleep’ vs prze-spać ‘oversleep’, za-spać ‘oversleep’, na-spać się ‘sleep long enough’, ode-spać ‘make up for one’s lack of sleep’.^[21] Whether this case has some theory-enriching potential remains a problem for future research.

Another group of verbs with interesting gaps in the distribution of prefixes are semelfactive verbs in Polish. Semelfactives, by and large, cannot be prefixed with perfectivizing prefixes of any kind, apart from the lexical spatial prefix od- (e.g. od-kopnąć ‘kick sth. away’, od-warknąć ‘rebuke’).

The very same class of verb vis à vis COS -ną- verbs in Czech and Polish has been analysed by Taraldsen Medová and Wiland (2019). The verbalizing suffix, common to both classes of verbs, is coded as NU in their Nanosyntactic research. They show convincingly that differences in the meaning of semelfactives and degree achievements (COS telic verbs) and in their morpho-syntactic properties can be satisfactorily coded without resorting to such theory-costly devices as zero morphemes. Although Taraldsen Medová and Wiland (2019) do not acknowledge the telic/atelic distinction characteristic of COS -ną- verbs, it might seem that an analysis of atelic/telic COS verbs can be conducted along similar lines, with telic verbs coding more complex Nanosyntactic verbal structures (including the structural layer of telicity) than atelic verbs, with the use of identical formal exponents (excluding the zero morpheme). However, the authors link the relative complexity of phonologically identical Nanosyntactic elements (NU in this particular case) with the relative complexity of categorial roots. Putting it very crudely, semelfactive (more complex) NU selects nominal, more complex roots, while the degree achievement NU selects less complex adjectival roots.

Quite apart from the fact that within DM roots are treated as a-categorial (especially Borer 2005a, b), so categorial differentiations cannot be applied to them, roots in our data are identical for atelic and telic verbs, unlike the roots of semelfactives and degree achievements in the Nanosyntactic analysis. Consequently, we could not claim that telic morphological elements subcategorise for some specific kind of roots, while atelic morphological elements are for some other type. The fact that we do not see a solution to the telicity problem within the Nanosyntactic Theory does not mean of course that such an analysis cannot be put forward.

As this short excursus shows, there are many problems tangential to the main interest of this paper, and many vistas open concerning the semantic properties of particular classes of verbs opposite their morpho-syntactic structure.

8 Conclusion

Relationships between structures proposed for particular classes of verbs resulting in general semantic properties of such classes vs (un)availability of prefixal subclasses are a very vast and little-researched area of morpho-syntax within DM. Hopefully, this article will contribute to the research in this area.

We have attempted in this article to explain in terms available in DM to account for the relationship between a gap in the prefixal system of Polish COS verbs and their dual atelic/telic properties. As the temporal modification test points to the existence of homonymous telic/atelic COS verbs in Polish, we maintain that the gaps in the distribution of prefixes with COS verbs are explicable in terms of structural distinctions proposed in (32) and (35). Telic COS verbs are equipped with the RP (32), while atelic ones have no such phrase in their morpho-syntactic structure (35). The head of the RP is filled with the zero-morpheme in the context of -e-/-ej-, -ną-, -ow-a- verbalizers, which pre-empties any possibility of filling the head with any other overt telicizing prefix. As atelic COS structures offer no RP to be filled with aspectual prefixal material, COS verbs never take overt lexical prefixes. The addition of pure perfectivizers and superlexical prefixes is unhindered in COS telic structures since they are added higher up in the morpho-syntactic structure.

Additionally, our results support a distinction between telicity and perfectivity postulated for Slavic languages in some sources (e.g. Smith 1991, Borik 2002, Filip 2003, 2008, among others) as in the case of COS verbs telic heads occupy very different positions in morpho-syntactic structures than pure perfectivizers and superlexical prefixes introducing perfectivity. Within the present model, COS verbs come in three flavours: imperfective telic and atelic, depending on whether they contain the RP in their structure, as well as perfective and telic, when prefixed with pure perfectivizers and superlexical prefixes. Consequently, the analysis supports the concepts that telicity and perfectivity must be held as separate language functions in Polish. Further research in this area is called for in other Slavic languages.

Abbreviations

ACC

accusative

COS

change of state

DAT

dative

DP

determiner phrase

DM

Distributed Morphology

GEN

genitive

I

inflectional morpheme

IMP

imperative

INF

infinitive

INS

instrumental

ITR

intransitive

IPFV

imperfective

L

lexical prefix

LOC

locative

NOM

nominative

PFV

perfective

PL

plural

PP

pure perfectivizer

PRS

present

PST

past

RP

result phrase

S

superlexical prefix

SG

singular

SI

secondary imperfective

TH

verbal theme-forming morpheme

TR

transitive