Verb-to-noun conversion in Polish: Multiple schemas in Construction Morphology

1 Introduction: conversion, zero-derivation, paradigmatic derivation

This paper aims to demonstrate how verb-to-noun conversion in Polish can be dealt with in the framework of Construction Morphology (Booij 2010; Masini and Audring 2019).

Conversion can be defined as “a derivational process linking lexemes of the same form but belonging to different word-classes” (Bauer and Valera 2005: 8). It is an affixless operation, distinct from affixation processes. Verb-to-noun conversion in English can be illustrated by the derivation of deverbal nouns, such as kick ‘an act of kicking’ and play ‘a piece of dramatic writing; a performance on the stage’.

The term “zero-derivation”, on the other hand, implies a derivational process which involves the addition of zero-affixes and which is parallel to word-formation processes that involve the addition of overt suffixes. According to Marchand (1969) and Kastovsky (2005), the occurrence of overt nominalizing suffixes, such as -ation and -ment, in the English deverbal action nouns modernization and development, provides support for the recognition of the nominalizing zero suffix in the English nouns change and return.

In the case of Polish, there exist a number of overt suffixes which derive deverbal nouns (so called “substantiva deverbalia”) functioning as names of actions, acts or processes, e. g. -k(a)^[1] in przeróbka ‘alteration, makeover’, -ek in upadek ‘fall’, -unek in rabunek ‘robbery’, -acj(a) in organizacja ‘organization’, -anin(a) in bieganina ‘hustle, comings and goings’, -eż in grabież ‘plunder’, -b(a) in prośba ‘request’. Moreover, verbal (gerundive) nominals can be formed in a regular way by attaching the suffix -ni(e)/-ci(e) to the verb stem, e. g. pisanie ‘writing’ from pisać ‘to write’ and wypicie ‘drinking up’ from wypić ‘to drink up’. The occurrence of those overt nominalizing suffixes can be used as a piece of evidence in favour of postulating zero suffixes in such deverbal action nouns in Polish as powrót ‘return, comeback’ (from powrócić ‘to return, to come back’) and zmiana ‘change’ (from zmienić ‘change’).

However, Polish morphologists (e. g. Waszakowa 1993; Grzegorczykowa and Puzynina 1998; Szymanek 2010) prefer to use the term “paradigmatic derivation”, instead of “zero-derivation”, when talking about an affixless category change. The so-called “paradigmatic formative” in the Polish nouns powrót ‘return, comeback’ and zmiana ‘change’ signals the change of the meaning and word class of derivational bases by means of changes in their inflectional paradigms.

In this paper (for the reasons discussed in the following sections), Polish verb-noun pairs, such as those in (1), are treated as conversion pairs. The noun is the converted mate since it is motivated semantically by the verb.

When providing verbs related to the conversion nouns in (1) and in the other sets of examples below I usually give the imperfective form for unprefixed verbs, such as budować ‘to build’ in (1c), and the perfective form for prefixed verbs, such as wybuchnąć ‘to explode’ in (1a). This is motivated by the fact that prefixed imperfective verbs, such as podpisywać ‘to sign’ and wybuchać ‘to explode”, are derived from corresponding prefixed perfective verbs by means of thematic (stem-forming) suffixes -iwa-/-ywa-, -owa-, -a- (see Szymanek 2010: 134 on Secondary Imperfectives in Polish). Sometimes, for expository purposes, I select the (perfective or imperfective) verb form which bears greater phonetic similarity to the noun, e. g. the prefixed imperfective verb rozpadać się ‘to disintegrate’, instead of the perfective verb rozpaść się ‘to disintegrate’, as the base for the noun rozpad ‘disintegration’ in (2b). I do not claim that the noun inherits the aspectual characterization of the imperfective or perfective verb form given in brackets. As pointed out by Rozwadowska (1997: 64) and Bloch-Trojnar (2013: 260–262), Polish deverbal nominals (including conversion nouns) are aspectually neutral.^[2] They differ in this respect from verbal (i. e. gerundive) -nie/-cie nominals, which preserve the aspect of related verbs.

Polish lexemes analysed in this paper appear in their citation forms: nouns are given in nom.sg case and verbs in their infinitival form.The infinitival ending -ć is immediately preceded in the verbs in (1) by an appropriate thematic suffix: -ną-, -a-, -owa- or -i-.The nouns in (1a), (1b), (1d) have no overt case marker, while in the nouns in (1c), (1e) the final vowel -a signals nom.sg case.

In some of the instances of verb-to-noun conversion under consideration there is no phonological modification of the stem, as in (1a–c). There is only a change of the syntactic category of the base (resulting in the change of the inflectional pattern of the derived word) and a change of its semantic interpretation. However, verb-to-noun conversion in Polish may be accompanied by internal modification^[3] (as shown in [1d–e]), e. g. the vocalic alternation /u/ – /ɔ/ in the pair nadzór ‘supervision’ – nadzorować ‘to supervise’, or the consonantal alternation /n/ – /ɲ/ co-occurring with the vocalic alternation /a/ – /ɛ/ in zmiana ‘change’ – zmienić ‘to change’.

In the sections to follow I will demonstrate that deverbal nouns in Polish derived by means of conversion/zero-derivation fall into several subsets of lexemes which are determined by their semantic interpretation and inflectional paradigms. The eventive and non-eventive readings of deverbal conversion nouns in Polish are discussed in Section 2. The inflectional properties of those nouns are analysed in the next section. Problems resulting from the proliferation of zero-affixes are also presented in Section 3. Section 4 shows how the proliferation problem can be handled in the framework of Construction Morphology. Section 5 contains conclusions.

2 The range of meanings of deverbal conversion nouns in Polish

The deverbal nouns analysed here include both names of events and names of people or objects involved in an event. As will be shown at greater length in Section 4, many deverbal conversion nouns in Polish are polysemous, exhibiting both eventive and non-eventive readings.

Event nouns can denote processes and actions, as in (2), or acts (i. e. episodes/instances of V-ing), as in (3).^[4] The conversion nouns in (2) can be treated as argument-supporting nominals (see Rozwadowska 1997). They can occur with internal arguments and (optionally realized) external arguments. They can also be accompanied by aspectual modifiers. This is shown by the phrases odbiór listu przez pozwanego ‘the receipt of the letter by the defendant’ and przekład czterostronicowego dokumentu w ciągu godziny ‘the translation of a four-page document within an hour’.

Grimshaw (1990), Rozwadowska (1997), Alexiadou and Grimshaw (2008), and Borer (2013), among others, draw the basic distinction between Event Nouns and Result Nouns (the latter also referred to as “Referential Nouns”). They point out that deverbal nouns (cross-linguistically) display the Event/Result polysemy, e. g. the Polish noun przekład ‘the process of translating; something that is translated’, as well as the English nouns building or invention.

However, it is useful to recognize more subtle distinctions among non-eventive interpretations of deverbal nouns. When describing the range of referential readings for Polish conversion nouns in this section, I will rely mainly on the classification of senses postulated for English deverbal nouns by Bauer et al. (2013: 207–212), mentioning additionally some semantic categories proposed for Italian suffixal nouns by Melloni (2011) or for English conversion nouns by Cetnarowska (1993) and Lieber (2016). I will also make reference to the discussion of actional and non-actional senses of Polish suffixal and affixless nominals by Waszakowa (1993), Grzegorczykowa and Puzynina (1998), Cetnarowska (1996) and Bloch-Trojnar (2013).

The classification of senses suggested by Bauer et al. (2013) for English nominals is adopted here for Polish nouns with some modifications.^[5] For instance, Bauer et al. (2013) identify the instrument reading, which is captured by means of the paraphrase ‘the thing that VERB-s, a way of VERB-ing’.^[6] I provide a slightly different paraphrase in (4) for a broad (supra)category named INSTRUMENT/INANIMATE_AGENT/MEANS, that is ‘the thing that VERB-s, something one can VERB with’. The term “non-animate agent” (or “Impersonal Agent”) is used by Booij (2010: 77–80) with reference to names of instruments which can be conceptualized as personified agents, e. g. printer ‘a machine that prints paper documents’. Deverbal nouns with the “means” interpretation are those which typically “make reference to to the material/object used to perform a specific action; but, importantly, this entity can also be interpreted as the object or artefact resulting from the accomplishment of the action” (Melloni 2011: 112), for instance decorazione ‘decoration’ and isolamento ‘insulation’ in Italian. Nouns paraphrasable as ‘the thing that VERB-s’ also include names of causers or stimuli of psychological states, such as Italian attrazione ‘attraction’ (Melloni 2011: 293). Basing on the aforementioned works, I adopt the taxonomy in (4) for Polish conversion nouns.

When discussing English denominal verbs which are formally identical with their bases, Štekauer (1996) observes that the recognition of such verbs as derived by means of zero-derivation would necessitate the postulation of a number of homonymous zero affixes (represented as -ø), each of them associated with a distinct semantic function,^[7] for instance -ø₁ which derives privative verbs (skin), -ø₂ which derives locative verbs (jail), -ø₃ which forms instrument verbs (hammer), and -ø₄ which derives similative verbs (ape).

A similar criticism can be formulated against the analysis of the Polish nouns given in (4) as zero-derivatives, as this would lead to the (undesirable) recognition of more than ten nominalizing zero-affixes, each resulting in a different semantic change. The zero-affix -ø₁ could, for instance, be postulated as the one deriving names of processes and actions (rozpad ‘disintegration’), -ø₂ as deriving episodic nominalizations (skok ‘jump’), -ø₃ deriving names of abstract results (podział ‘division’), -ø₄ deriving names of products (napis ‘inscription’), -ø₅ deriving names of instruments/inanimate agents (wyciąg ‘hoist’), -ø₆ deriving agentive nouns (rząd ‘government’), -ø₇ deriving locative nouns (właz ‘inlet’) etc.

Such an objection may not be directly applicable to syntax-based approaches to word-formation, e. g. the Exoskeletal model (Borer 2013) and Distributed Morphology (DM) model (Alexiadou et al. 2010, Iordăchioaia 2020). In such approaches distinct modes of deriving are postulated for Event Nominals and for Referential Nominals. While Event Nominals contain verbal projections in their structural representations, Referential Nominals (such as English drive ‘a private road, driveway’ or Polish napis ‘inscription’) are treated – in DM accounts – as root derivatives, which contain category-less roots, lack any verbalizers and therefore are not regarded as deverbal formations.^[8] Since differences between various non-eventive readings are largely neglected in syntax-based approaches (though see Iordăchioaia et al. 2020), I will focus here on lexicalist approaches to zero-derivation, such as Construction Morphology.

As is shown in the next section, deverbal nouns derived by means of conversion/zero-derivation in Polish belong to various inflectional classes, which further complicates their description.

3 Inflectional properties of Polish conversion nouns

Polish is characterized by a grammatical gender system, in which three basic grammatical genders are recognized: masculine, feminine and neuter. Masculine gender can be divided into three (sub)genders (see Orzechowska 1998: 272): virile (i. e. masculine personal, such as mąż ‘husband’), masculine animate (kot ‘cat’), and masculine inanimate (płot ‘fence’). Nouns belonging to those three masculine (sub)genders usually take a phonologically null (i. e. zero) ending as the marker of nom.sg. They differ in their patterns of syncretisms in the inflectional paradigm, e. g. masculine personal and masculine animate nouns exhibit the syncretism of gen.sg and acc.sg, while for masculine inanimate nouns the syncretism involves nom.sg and acc.sg case forms.

The deverbal conversion nouns mentioned in (3) in the previous section belong to masculine nouns, e. g. the masculine personal noun szpieg ‘spy’ and the masculine inanimate noun wyciąg ‘hoist’. Deverbal nouns of the masculine gender represent the largest group of conversion nouns.^[9] There are also nouns resulting from verb-to-noun conversion which exhibit feminine grammatical gender. At least two inflectional classes can be distinguished in the case of feminine action nouns.^[10] The majority of them, referred to as hard-stem feminine nouns, take -a as the marker of nom.sg, e. g. obron-a ‘defence’ (from obronić ‘to defend’). A smaller subgroup of feminine gender nouns consists of nouns which take the phonologically null inflectional ending in the nom.sg case, e. g. pogoń ‘chase’ (from pogonić ‘to chase’). Nouns in the latter group are referred to as soft-stem nouns.^[11] The final consonant of their stems is either a prepalatal consonant, e. g. /ɲ/ in pogoń ‘chase’ and // in odpowiedź ‘reply’ (from odpowiadać ‘to reply’), or one of the so-called “morphophonologically soft” consonants (including the lateral /l/, the retroflex fricatives /ʂ/, /ʐ/, the retroflex affricates //, // and the alveolar affricates //, //), e. g. // in pomoc ‘help’ and /l/ in myśl ‘thought’ (Strutyński 1996: 145–147).

Feminine hard-stem nouns exhibit various actional and non-actional senses, as shown in (5) and (6) below.

Feminine soft-stem conversion nouns are less common than feminine hard-stem conversion nouns. They occur as names of events (7a) or exhibit non-eventive senses (7b–e).

There are very few deverbal conversion nouns which show neuter gender and take -e as the marker of nom.sg.^[12] They exhibit mainly the sense of instrument or location, e. g. złącze ‘connector’ (cf. złączyć ‘to connect’), przyłącze ‘(gas or power) connection’ (cf. przyłączyć ‘to attach’) and osiedle ‘housing estate’ (cf. osiedlić ‘to settle’).

If we disregard the occurrence of neuter gender affixless nouns and attempt to postulate distinct WFRs (or distinct zero affixes) for feminine gender and masculine gender nominals (exhibiting various meanings), this might suggest the need for a recognition of over twenty^[13] zero affixes, each of them associated with a distinct semantic and inflectional specification^[14] of resulting conversion nouns. For instance, the derivation of deverbal conversion nouns with the event reading (Action/Process/Act), or with the agentive reading, would call for the distinct zero affixes postulated in (8)–(9).^[15]

In the next section the question will be posed how the problem of proliferation of zero affixes (resulting from the need to account for various types of deverbal affixless nouns) can be dealt with by proponents of Construction Morphology.

4 A Construction Morphology account

The basic units of analysis in Construction Morphology are constructions, which are defined as conventionalized pairings of form and meaning. Construction schemas consist of statements of morphosyntactic and phonological operations accompanied by the specification of the meaning of constructions. The schema for noun-to-verb conversion in English, as exemplified by the denominal verbs book ‘to reserve (accommodation etc.) in advance’ and scholarship ‘to attend an institution on a scholarship’, is given in (10a), where the subscripts i, j are indices, and where α and β stand for grammatical gender and/or inflectional classes. The lower case letter x is a variable standing for some phonological material, e. g. [bʊk]. The left-hand part of the schema in (11a) says that nouns are converted into verbs without the addition of any affix. Zero affixes are signs without form, therefore they are problematic in sign-based theoretical frameworks, such as the theory of Construction Morphology. The right-hand part of (10a), following the arrow, states the meaning of the converted verb: it can denote an event which is related in some way to what is denoted by the noun (SEM_i). The double headed arrow signals the association between both parts of the schema.

When modified to account for English verb-to-noun conversion, (10a) can be rewritten as (10b). The output noun is a name of an event which is denoted by the base verb (SEM_i), e. g. the event of walking in the case of [[walk]_V]_N.

Since deverbal conversion nouns exhibit various actional and non-actional senses (Waszakowa 1993), and since Polish deverbal conversion nouns belong to three genders (and to various inflectional classes), as shown in Sections 2 and 3, multiple construction schemas need to be postulated to model this type of conversion.

In the hierarchical constructicon, constructions with a higher level of generality (i. e. a higher level of abstractness) are instantiated by more specific constructions, which (by default) inherit the properties of the higher-level constructions (unless they clash with some features of the lower-level construction). The general schema in (10b) corresponds to the three more specific schemas in (11), in which the variables α and β have been replaced by appropriate formal features, signalling the gender and the inflectional class of the action nouns in question.^[16]

As stated by Booij (2010) or Masini and Audring (2019), a morphological schema is constructed on the basis of existing complex words. It has a declarative character and it motivates the internal structure of such words. Appropriate examples of Polish conversion nouns which serve as instantiations of the relevant construction schemas in (11) below are provided in brackets.^[17]

The variable x used for the phonological form in (11a–c), as well as in further construction schemas below, implies the identity of the phonological form of the base and the phonological form of the derived noun.^[18] It needs to be clarified that this identity refers to the stem of the derived affixless noun,^[19] e. g. obsług- in obsługa ‘servicing’, and the basic verb stem (or the extended root). The latter term refers to the part of the verb which remains once the thematic suffix and the inflectional ending are removed, i. e. obsług- in the verb obsługiwać (ipfv) ‘to wait on somebody’, where -iwa- is the thematic (Secondary Imperfective) suffix and -ć is the infinitival ending. The basic verb stem consists here of the prefix ob- and the root sług- ‘serve’.

A question could be asked if the schemas in (11a–c) are appropriate for stating the relationship between verbs and affixless deverbal nouns whose stems exhibit morphophonological alternations (as mentioned in Section 1). In the model of Construction Morphology (Booij, 2010; 2019), so-called second order schemas (also referred to as “sister schemas”) are able to express the paradigmatic relation (represented by ≈) between stem allomorphs. The second order schema which accounts for /u/ – /ɔ/ alternation could roughly take the form of (12).^[20] This schema can be employed to state the relationship between the stems in the noun-verb pair (1d), i. e. nadzór ‘supervision’ and nadzorować (ipfv) ‘to supervise’ (where -owa- is the thematic suffix and -ć is the infinitive marker).

The occurrence of the vocalic /a/ – /ɛ/ alternation attested in the noun-verb pair in (1e), i. e. zmiana ‘change’ and zmienić (pfv) ‘to change’, is stated in the schema in (13).

Further construction schemas, apart from those in (11), are required to account for the inflectional classes and the semantic interpretation of deverbal conversion nouns which exhibit non-actional readings in Polish. For example, the schemas in (14) represent the inflectional properties and the (general) meaning of agentive nouns in Polish derived by means of verb-to-noun conversion.

The schemas in (15) generalize over the existing deverbal conversion nouns in Polish with the instrument reading (paraphrasable as ‘something that V-s, something one can V with’).

At first glance, it seems that postulating multiple construction schemas (to account for various types of deverbal conversion nouns in Polish) is as undesirable and costly as recognizing multiple zero affixes. However, it can be argued that having over twenty distinct construction schemas for verb-to-nouns conversion in Polish comes at a lower cost than having twenty four or more homonymous (and unrelated) zero affixes. This is because construction schemas form networks of relations, in which one (fairly) abstract schema can be instantiated by several more specific schemas.^[21] This was shown above for the general schema in (10b) and its instantiations in (11). The three specific, hence lower-level, agentive schemas in (14) can be treated as subschemas for (16a), while the schemas in (15) – as subschemas for (16b).

The schemas for Agents and Instruments in (16a) and (16b) can be treated as instantiations of the schema in (17). The variable v in (17) stands for some phonological material, e. g. the basic verb stem nadzor(ować) ‘to supervise’, while x and y represent the external and the internal argument of the verb. The schema in (17) says that a verb converted into a noun denotes the external (x) argument^[22] of the verb.

Schema (17) generalizes over conversion nouns with the subject-type interpretation, i. e. those which denote personal (human) agents, inanimate agents, instruments, means and causers. The possibility of various subject-type readings with a given class of morphologically complex words is observed for suffixal nouns, such as -er nouns in Dutch and English, which can denote personal agents, inanimate agents or instruments (Booij 2010; Bauer et al. 2013). Polish conversion nouns may be ambiguous in a similar way, e. g. ochrona ‘people who provide protection, bodyguards’ (Collective Agent) and ‘something that provides protection’ (Inanimate Agent) (as in ochrona do ładowarki telefonu ‘protect bumper for a phone charger’). This ambiguity is predictable given the higher-level schema in (17) and its lower-level instantiations in (16).

Schema (17) is, in turn, dominated by schema (18), which exhibits a high level of generality and predicts that deverbal conversion nouns can refer to conceptual^[23] participants in particular events (including, among others, instruments, agents, themes/patients, and results).

The schema in (19) is even more general and dominates the schemas in (18) and (10b). It says that deverbal conversion nouns can either function as names of events or as names of conceptual participants in those events.

The general schema in (19) helps predict the logical (i. e. regular) polysemy^[24] of deverbal conversion nouns. They can denote events, abstract results of those events or they can show various referential senses. Relevant examples of the regular polysemy of Polish conversion nouns are given in (20) and are also provided by Waszakowa (1993) or Bloch-Trojnar (2013).

Booij (2018: 7) observes that “[i]t is important to be able to express generalizations about complex words on different levels of abstraction, since a set of complex words may consist of subsets with properties of their own.” For instance, Polish masculine gender conversion nouns may refer to groups of people who perform a given action (such as zarząd ‘management’ and nadzór ‘supervision’), yet they tend to denote individual agents, such as zbieg ‘fugitive’ and szpieg ‘spy’ (Waszakowa 1993: 50–52). Polish feminine gender conversion nouns with the agentive reading usually denote collective agents, as is shown in (21) below (see also Waszakowa 1993: 76).

Masini and Audring (2019: 387) argue that the postulation of numerous related schemas in a hierarchical constructicon is an advantage of the machinery of Construction Morphology (and Construction Grammar, in general). Such lower-level schemas share some of their properties (which are inherited from the more abstract schema) but they differ in their degree of productivity. Some schemas are used for the description of non-productive processes, such as the conversion of verbs in Polish into feminine gender soft-stem nouns. Others represent either mildly productive or very productive patterns of verb-to-noun conversion.^[25] The schema in (11a) is very productive, as it can give rise to neologisms (attested in informal Polish) which are masculine gender conversion nouns with the event reading,^[26] such as those in (22).

Jadacka (2001: 91) provides some examples of novel masculine conversion nouns in Polish (formed at the end of the twentieth century) which exhibit non-eventive readings only, e. g. the agentive noun przetłumacz ‘bad translator’ (cf. przetłumaczyć ‘to translate’),^[27] or names of Themes/Patients, such as wszczep ‘implant’ (cf. wszczepić ‘to implant’). More recent examples of agentive or patientive expressive neologisms of this type, found by me in the dictionary of Polish urban slang at http://miejski.pl, are given in (23).

The feminine gender paradigmatic formative which derives hard-stem nouns is treated by Waszakowa (1993) as exhibiting weak productivity, yet some slang words of this type have been derived recently in Polish, as is shown in (24) (see http://miejski.pl).

Consequently, productive construction schemas in CxM, such as the schemas for masculine gender conversion nouns in Polish, function not only as redundancy statements (which generalize over institutionalized words) but can be used as sets of instructions for deriving novel instantiations of those constructions.

5 Summary and conclusions

The examination of deverbal nouns derived in Polish by means of an affixless morphological operation (which can in principle be regarded either as conversion or zero-derivation) has shown that they belong to several inflectional classes. The majority of them are masculine gender nouns which take a phonologically null inflectional ending as the marker of nom.sg, e. g. nadzór ‘supervision’ and rzut ‘throw’. They represent a very productive pattern, which can be employed to derive novel deverbal nouns. A sizeable group of feminine gender hard-stem nouns, with the vocalic ending -a as the marker of nom.sg (e. g. zmiana ‘change’ and obrona ‘defence’) instantiate another pattern of conversion, which can be regarded as mildly productive but which can give rise to neologisms. A smaller group of deverbal affixless derivatives contains feminine gender soft-stem nouns, with a phonologically null inflectional ending in nom.sg, e. g. pomoc ‘help’ and więź ‘tie, bond’. Finally, there are very few deverbal affixless nouns which are inflected as neuter gender nouns and take the vocalic ending -e or -o as the marker of nom.sg, e. g. łącze ‘link’. Deverbal conversion nouns which belong to the first three inflectional types exhibit the range of meanings characteristic of derived nominals (cross-linguistically). Apart from denoting events, they allow non-eventive interpretations, such as the agentive reading, the instrument reading, the sense of the Theme/Patient, Product, Location etc. Neuter gender conversion nouns are found in the instrument or locative sense. If we assumed that each zero-suffix which is employed in coining Polish deverbal nouns should specify the inflectional properties as well as the semantic interpretation of conversion nouns, it would be necessary to distinguish over twenty nominalizing zero suffixes. In a theoretical framework in which the morphological process involved in coining the nouns under analysis is treated as the operation of conversion (and not as zero-affixation), a similar problem may arise. In the theory of Construction Morphology abstract morphological schemas are constructed on the basis of existing morphologically complex words. Separate morphological schemas are required for various classes of Polish deverbal conversion nouns with distinct inflectional characteristics and different semantic interpretation. However, this proliferation of schemas which model verb-to-noun conversion in Polish has some advantages. In the hierarchical lexicon assumed in Construction Morphology, morphological schemas exhibit different levels of generality. They form hierarchical networks, in which several more specific (low-level) schemas may be dominated by a single general schema. This assumption provides internal structure to the lexicon. Low-level schemas dominated by a particular higher-level schema may differ in their degree of productivity. Productive schemas for masculine gender nouns (in eventive and non-eventive senses), which can serve as patterns for new formations, contrast with unproductive schemas for feminine gender soft-stem nouns, which only generalize over existing deverbal nouns of this type.

The most general (i. e. abstract) schema which can be posited for deverbal (affixless and suffixal) nouns states that a derived noun can either serve as a name of a given event or as a name of a participant involved in the event. This type of schema helps predict the patterns of regular polysemy displayed by deverbal nouns.

An additional advantage of the account of verb-to-noun conversion presented in this paper in the model of Construction Morphology is the use of second-order schemas (i. e. sister schemas) which state a paradigmatic relation between stem allomorphs attested in verb forms and in deverbal nouns (e. g. zmiana ‘change’ and zmienić ‘to change’).