PART 1: DIACHRONIC EVIDENCE

 

The rise of the WZIAC (TAKE) Serial Verb Construction in Polish

 

 

Alexander Andrason^I; Malgorzata Gebka-Wolak^II; Andrzej Moroz^III

^IFaculty of Humanities, University of Cape Town, South Africa. E-mail: aleksand@hi.is
^IIDepartment of Polish Language, Nicolaus Copernicus University in Torun, Poland. E-mail: mge@umk.pl
^IIIDepartment of Polish Language, Nicolaus Copernicus University in Torun, Poland. E-mail: amoroz@umk.pl

 

 

---

ABSTRACT

The present study is dedicated to the emergence of an asymmetrical serial verb construction (SVC) with the verb wziqc in Polish. By making use of a dynamic prototype-driven approach to linguistic categorization and by reviewing the historical corpora that range from the first Old Polish texts in the 14^th c. until the end of the New Polish period in 1939, the authors conclude that the wziqc SVC has resulted from the fusion of the original conjunctively coordinated (CC) clauses. Although two types of clause-fusion mechanisms have operated during the grammaticalization of the wziqc SVCs, their contribution to this process has been dissimilar. The evolution from the syndetic CC with the coordinator i to the wziqc SVC via a pseudo-coordinated (PC) stage (i.e., the wziqc-i PC) has constituted a faster and stronger drift, while the more direct evolution originating in the asyndetic CC with wziqc has been slower and less pervasive.

Keywords: Serial verb constructions; take verbs; grammaticalization; Polish

---

 

 

1. Introduction

The present article studies the emergence of a serialized construction built around the minor (and construction-first, i.e., Vi) verb wziqc lit. 'take' in Polish. Specifically, we verify whether this serial verb construction (SVC) results from the fusion of conjunctive coordinated (CC) clauses, syndetic and/or asyndetic, which constitute the two most likely scenarios that have been hypothesized in scholarly literature (Andrason 2018a; 2018a; 2019b). To do so, we trace the attestations of the various constructions built around the verb wziqc - whether coordinated (syndetic and asyndetic), pseudo-coordinated (the so-called wziqc-i PC, lit. 'take and'), or serialized (the wziqc SVC) - from the earliest texts in the 14^th c. until the end of the so-called New Polish period in 1939. In our analysis we make use of a dynamic, prototype-driven approach to linguistic categorization (Andrason 2016a) and thus the three categories distinguished, i.e., conjunctive coordination, pseudo-coordination, and serial verb construction (Haspelmath 2004; 2007; Aikhenvald 2006; 2011; 2018; Dixon 2006; and Andrason 2018a; 2019a; 2019b).

In order to achieve our goal, we structure the article in the following manner: in section 2, we familiarize the reader with the previous hypotheses regarding the origin of the wziqc SVC and explain the theoretical framework underlying our research. In section 3, we introduce original diachronic evidence. In section 4, we evaluate this evidence within the adopted framework and answer the research question. Section 5 concludes the study.

 

2. Background

2.1 Literature review

The origin of the wziqc SVC has not been researched in detail.¹ The only studies that overtly deal with this issue draw their conclusions from a diachronic interpretation of synchronic data (Andrason 2018a; 2019a) by making use of a method commonly referred to as 'dynamization of synchrony' (cf. Croft 2003; Haspelmath 2003). Additionally, possible sources of the wziqc SVC may be inferred by analogy to the origins proposed in a study dedicated to another, more or less related, construction: the pójsc SVC (Andrason 2018b). As is the case of research on the wziqc SVC itself, the study discussing this other construction fails to support its hypotheses with direct diachronic data, instead drawing on synchronic evidence (Andrason 2018b).

In the concluding parts of his articles dedicated to the wziqc SVC (2018a) and the wziqc-i PC (2019a), Andrason proposes the following: the wziqc SVC derives from the wziqc-i PC, which itself descends from a genuine syndetic CC pattern with the verb wziqc used as the predicate in the first conjunct clause (Andrason 2019a:185-186; see also Andrason 2018a:606; 622-623). Andrason supports his proposal by four, closely related arguments: the synchronous presence of the three construction types built around the verb wziqc in modern Polish (i.e., CC, PC, and SVC; Andrason 2018a; 2019a); the "plottability" of these three variants into a coherent semantic map and the widely accepted diachronic foundation of such maps (cf. Haspelmath 2003; Janda 2015; Andrason 2016a); the soundness of the diachronic path regulating the life of coordination-based SVCs proposed in typological studies, according to which the map of the three wziqc constructions has been plotted (cf. Aikhenvald 2011; 2018; see section 2.2 below); and the robust, cross-linguistically pervasive principles of grammaticalization that explain the changes observed in the constructions located in different sections of this map (cf. Hopper & Traugott 2003). Under this proposal, a bi-verbal, bi-event, bi-clausal, and bi-predicative structure - i.e., two clauses coordinated by a canonical coordinator i 'and' - has enhanced its cohesiveness by acquiring a mono-event, mono-predicative, and mono-clausal profile (Andrason 2019a:186). As the element i has been preserved, although not as a coordinator but rather as a "dummy" junction, the resulting structure is a PC.² Subsequently, with the further increase of cohesiveness and the loss of the juncture i, the construction has acquired the status of a canonical SVC (for details see Andrason 2018a:606, 611, 622-623; 2019a:184-186).

An alternative origin of serial verbs in Polish is suggested by Andrason (2018b) in relation to the pójsc SVC, i.e., a serializing pattern built around the perfective verbpójsc 'go' as its minor verb. By extension, this hypothesis could also apply to the wziqc SVC. To be exact, the pojsc SVC may have originated from an asyndetic CC structure. It would thus derive from constructions in which the first conjunct, i.e., the one containing the verb pojsc, was followed by the second verbal conjunct without the intermediacy of any linking element (Andrason 2018b:41-43). Of course, the pseudo-coordinated variant (i.e., the pójsc-i PC) would, as in the proposal discussed in the paragraphs above, descend from the syndetic CC with i. As was the case of the wziqc SVC, Andrason arrives at his hypothesis by drawing on the dynamic map of the constructions built around the verb pójsc (i.e., the pójsc SVC, pójsc-i PC, and genuine CC with pójsc) and thus variation in uses and forms attested synchronically in modern Polish.³

Additionally, in his 2019 study, in two footnotes at the end of the article, Andrason considers another scenario, namely the development of the wziqc-i PC from the wziqc SVC through the introduction of the junction i by analogy to other pseudo-coordinated constructions built around motion verbs. This scenario, however, is, in Andrason's (2019a:186) view, implausible. First, it would violate the principle of unidirectionality of grammaticalization: that is, a construction that corresponds to a more advanced grammaticalization stage (SVC) - both morphologically and syntactically - would give rise to a construction reflecting a lesser extent of grammaticalization (PC). Second, the hypothesized evolution would violate the connectivity principle of the synchronic mapping: at certain times in the history of the wziqc constructions, the variants compatible with the initial stage (CC) and the final stage (SVC) would have been grammatical and productive, whereas the variant corresponding to the intermediate stage (PC) would have been ungrammatical. Third, the PC and the SVC built around the verb wziqc are the most common and the most grammaticalized among all types of pseudo-coordinated and serialized constructions found in Polish (see Andrason 2018a; 2018b; Gebka-Wolak & Moroz 2021). Therefore, "it is unlikely that the juncture-free patterns would develop a juncture under the pressure of other, less common, less productive, and more restricted serializing patterns" (Andrason 2019a:186).

To conclude, out of the three hypotheses presented above, only the first two are, in our view, plausible. Both imply a clause-fusion mechanism as the source of verbal serialization: (a) syndetic CC with wziqc in the first clause and the coordinator i linking two verbal conjuncts -> wziqc-i PC -> wziqc SVC; (b) asyndetic CC with wziqc in the first clause -> wziqc SVC. Our study will determine whether the wziqc SVC has indeed emerged from clause fusion and which one - if any - of the two specific clause-fusion mechanisms proposed is more probable in light of direct diachronic evidence. However, before presenting the results of our empirical research, we will explain in detail the framework that underlies our study.

2.2 Framework

As explained in the introduction section, in order to determine the source of the wziqc SVC and, in particular, to verify whether this grammatical structure results from clause fusion, we will trace the attestations of the various constructions built around the verb wziqc - whether coordinated (syndetic and asyndetic), pseudo-coordinated, or serialized - in the available Polish texts, from the Old Polish period in the 14^th- c. until the end of the so-called New Polish period in 1939. The critical task in our empirical research thus consists of classifying the various uses of wziqc as instantiations of one of the three aforementioned construction types. In this classification, we use a prototype-driven approach to linguistic categorization. Accordingly, we understand each of the three categories as networks of more or less canonical members organized around ideal prototypes. The prototype itself - the conceptual nucleus of each category - is defined cumulatively as a set of cross-linguistically common and cognitively salient properties.

With regard to the prototype of (clausal) CC, the works presented by Johannessen (1998), Haspelmath (2004; 2007), and Andrason (2019a; 2019b) suggest the following: semantically, CC is a relatively straightforward combination of the formative parts, its meaning thus being compositional; the conjunctive coordinator used has a force equivalent to the operator λ in Classical Logic and η in Set Theory; conjuncts are separated intonationally (e.g. by a break, pause, or contouring); the verbs found in each conjunct exhibits their own and potentially distinct argument structure - this means that each conjoined verb may project its own subject and object(s) and that the adjuncts used may operate over a single verbal conjunct only; the verbs used in the conjuncts express independent events (e.g. consecutive or simultaneous) that need not temporally and spatially overlap - as a result, the verbs can host different, even opposite, TAM markers (this separate eventhood of each verb may be made overt with operators of time, space, manner/means, and instrument binding each verb individually); the conjuncts may have distinct polarity values and, when negated, each of them is marked by its own negative morpheme; the order of the conjuncts may be reversed "with no implications for the truth conditions of the sentence" (Andrason 2019a:166);⁴ the extraction of the argument or adjunct elements governed by the second conjunct and their placement in front of the first conjunct is ungrammatical unless these elements are also governed by the verb used in this first clause; conjuncts may be modified by the quantifier 'both' and/or coordinated bi-syndetically, conveying the sense equivalent to 'both...and' in English; when the CC sequence consists of more than two conjuncts, "all coordinators can be omitted with the exception of the last one" (ibid.); lastly, it is syntactically possible to replace the coordinator used in the construction by other coordinators found in the same language, whether conjunctive, disjunctive, or contrastive. The prototype of asyndetic CC exhibits identical properties with the exception of those related to the coordinator, since this element is absent.

With regard to the SVC prototype, the seminal books and chapters authored by Aikhenvald (2006; 2011; 2018), as well as works published by Muysken & Veenstra (1994), Dixon (2006), Bisang (2009), and Andrason (2018a; 2019a), allow us to discern the following key properties: similar to CCs, an SVC makes use of two finite verbs that can be employed as independent lexical verbs outside the SVC itself;⁵ however, in contrast to CC, an SVC is not a direct combination of the two verbs used but instead exhibits its own constructional meaning; the two verbs are not linked by markers of syntactic dependency - this thus precludes the use of conjunctive coordinators; the verbs are not marked by intonational phrasing and are not separated by pause and/or contouring; duplicate roles that would be projected by each verb separately are disallowed - instead, the construction exhibits a unitary argument structure, the verbs sharing their external (subject) and/or internal (objects) arguments, as well as all the adjuncts; the construction expresses a unitary event - the two verbs fail to be marked for incompatible TAM categories, instead being jointly bounded by the operators of time, space, manner/means, and instrument; polarity operates over the entire SVC rather than over a verb individually - therefore, a single negator is sufficient to negate the entire construction; the order of the verbs may not be reversed; the extraction of any argument or adjunct and their placement in front of the first verb is grammatical; two verbs occupy an adjacent position.⁶

As discussed by Johannessen (1998), De Vos (2004; 2005), Andrason (2019a), and Giusti, Di Caro & Ross (2022), the prototype of PC exhibits features associated with both the prototype of CCs and that of an SVC. One the one hand, PC complies with most properties typical of SVCs enumerated above with one important exception: PC contains an element that is identical (or highly similar) to a conjunctive coordinator used in genuine CC structures in the language. This property formally approximates PC to CC. However, contrary to CC, the linking element found in PC does not function similarly to the operators λ or n. It is rather a "dummy" or "empty" marker - the so-called junction or linker. Therefore, in further contrast with CC, the verbs used in PC cannot be modified by the quantifier 'both', nor can they convey the sense equivalent to 'both.. .and' if the first verb is headed by a genuine conjunctive coordinator.⁷

To summarize: (a) prototypical CC is a bi-event, bi-predicative, and bi-clausal; (b) prototypical PC and SVC are both mono-event, mono-predicative, and mono-clausal; (c) prototypical PC contains a junction formally identical to a conjunctive coordinator found in CC but absent in SVCs. Although the three prototypes may be instantiated in specific languages in a canonical manner, i.e., by constructions that comply with them fully, very often this compliance is partial. Indeed, many language-specific constructions exhibit intermediate and, thus, fuzzy profiles. Such intermediate profiles exhibited by instantiations of the three categories are comprehensible, in fact expected and necessary, if the three prototypes are understood as stages located along a single grammaticalization path - the clause-fusion path.

As proposed by Aikhenvald (2018:195), one of the main mechanisms leading to SVCs is clause fusion, whereby bi-clausal verbal structures gradually acquire a mono-clausal status.⁸ This typically involves the elimination of elements that have originally marked one of the verbal clauses as dependent on the other (ibid. 197, 199). Such dependency markers may be subordinizers, converbal suffixes, or other (inflectional) morphemes, as well as coordinating conjunctions - the most relevant for our study (ibid. 196, 199, 299). Accordingly, some SVCs emerge from original bi-clausal CC structures in which the two conjunct clauses were linked syndetically, i.e., by a conjunctive coordinator. Before becoming canonical SVCs, these biverbal sequences pass through the stage of PC (Bowern 2008:169; Hilpert & Koops 2008). At this stage, the original coordinator, the "erstwhile marker of dependency between two verbs" (Aikhenvald 2018:199), no longer indicates dependency but is rather used as "a 'dummy' linker" or "pure construction marker" (ibid.) - a pseudo-coordinator.⁹ As the development advances, the pseudo-coordinating junction tends to be reduced phonologically and semantically, ultimately being lost entirely. That is, its form may be radically distinct from the coordinator used in canonical CC (from which it has evolved) and/or the use of the original coordinator is no longer attested outside the bi-verbal (PC) construction itself (where it is, of course, no longer employed as a coordinator sensu stricto) (ibid.).¹⁰ This development is represented graphically below:

 

Figure 1

 

The clause-fusion mechanism proposed by Aikhenvald (2018) also makes room for a developmental scenario in which SVCs emerge from juxtaposition, including asyndetic CC, instead of the syndetic type discussed above (see Bowern 2008:171; Bisang 2009:802, 809; Palancar 2012:28).¹² The development of SVCs from asyndetic CC patterns follows virtually the same rules as the evolution originating from syndetic CC. The only significant exception is the cluster of changes that affect overt dependency markers, in particular conjunctive coordinators and thus the absence of the stage of PC (see Figure 2 below). Nevertheless, as in the evolution discussed above and as is typical of all grammaticalization processes, the progression along the path - again involving the change from bi-clausality and bi-eventhood to mono-clausality and mono-eventhood - is gradual and allows for a gamut of intermediate and fuzzy stages and thus constructions.

 

3. Evidence

3.1 Empirical research

As is typical of all studies that concern earlier stages of a language, one cannot make use of their own linguistic competence but is instead limited to the examination of historical and/or diachronic corpora.¹³ Due to the fragmentary and scattered nature of the historical material of the Polish language currently available, we adopted a deductive (rather than inductive) and heterogenous (rather than homogenous) procedure during our empirical historical-diachronic-corpus research. We did not start from observations and subsequently, through induction, formulated some generalizations but rather tested the hypotheses developed deductively (see section 2) and looked for a set of expected phenomena (cf. Kleszczowa 1991:91-92). Furthermore, we used dissimilar methods for searches effected in different corpora from different time periods instead of examining a single comprehensive corpus in a unitary manner.

The main idea guiding this non-uniform empirical work was twofold. We aimed to find all possible cases of SVCs and PCs with the verb wziqc from the time of the earliest records of the Polish language in the 14^th c. to the end of the New Polish period in 1939 (which coincides with the beginning of World War II). Alternatively, in case of the absence of wziqc SVCs and PCs in corpora from earlier stages of the history of the Polish language, we aimed to find the most similar constructions, namely, asyndetic and syndetic coordinated constructions in which the verb wziqc and another verb appear in the adjacent clauses and are inflected in the same tense, aspect, and mood (TAM) categories as well as person and number. Therefore, although our corpora and search methods were heterogenous, the syntax of questions used in searches was highly similar. That is, different search engines (see below) were programmed to retrieve sequences composed of two verbs, the first of which was one of the forms of the lexeme wziqc. Both verbs were expected to be inflected in the same tense, aspect, or mood, specifically: imperative, past tense, non-past tense (i.e., perfective present, which functions as a future tense in Polish), and infinitive.¹⁴ We also accepted three types of linear arrangements: one contiguous, whereby V1 and V2 appear next to each other, and two non-contiguous, specifically, those where V1 and V2 are separated by one and two segments.¹⁵ We allowed V1 and V2 to be separated by a conjunction (whether true or empty), comma, and other orthographic symbols, as well as combinations of all such "separatrices".

Traditionally, a historical corpus was an analogue collection of texts from which the researcher would manually extract examples relevant to their lexico-grammatical analysis. Currently, an increasing number of historical texts and, thus, corpora are being digitalized and made available in the form of electronic repositories with texts annotated and prepared for, more or less sophisticated, searches. Despite this, no comprehensive historical-diachronic electronic corpus of Polish has been developed thus far - there is no corpus of Old and pre-Modern Polish digital texts that would comply with representativity requirements and, at same time, allow for automated searches. This stems from two main reasons. First, the above-mentioned scarcity of extant linguistic material. Second, the changing rules of spelling conventions and grammar systems found in the various corpora already available, which render the transcription of texts - necessary for their digitalization - extremely difficult, time consuming, and costly (Górski, Król & Eder 2019: 12-14). Overall, we have made use of the following corpora in our empirical research, here arranged chronologically: Digital Corpus of Old Polish Text till 1500 (https://iip.pan.pl/publikacie-i-materialy/zasoby/korpus-tekstow-staropolskich), Corpus of 16^th c. Old Polish and a digital repository of texts from the so-called canonical sources of the Dictionary of Polish of the 16^th c. (https://spxvi.edu.pl/korpus/teksty), Corpus of Polish Texts from the 17^th and 18^th c. (https://korba.edu.pl), usually referred to as the Baroque corpus "KorBa", as well as the National Corpus of the Polish Language (http://nkjp.pl), which, albeit essentially a contemporary corpus (20^th and 21^st c.), also contains some older sources. The above corpora, available to all researchers, have been complemented with our own original corpus prepared from digitalized 19^th and pre-war 20^th c. texts.

Contrary to their chronology and the order presented above, in our empirical research, we proceeded from corpora that are more contemporary to those that are gradually more remote from today's Polish, i.e., from the 21^st-20^th c., through the 19^th-early 20^th c., 17^th-18^th c., and 16^th c., until the 14^th-15^th c.

The starting point of our research was the corpus of modern Polish, mainly from the 21^st and 20^th c. To extract the wziqc SVC and PC, we used two automatized search engines: Poliqarp (http://nkip.pl/poliqarp), which operates over the National Corpus of the Polish Language (Przepiórkowski, Banko, Górski & Lewandowska-Tomaszczyk 2012) and Monco (http://monco.frazeo.pl), which operates over RSS channels made available by more than 1500 news services (Pezik 2020:134). The main advantage of these two search engines is their ability to work on an immense number of texts. Poliqarp searches a database containing 2 milliard segments, while Monco searches 7 milliard segments. Both search engines include texts of considerable stylistic variability, which increases the probability of encountering relevant constructions, in our case, an SVC and PC with wziqc. Both engines also have a sophisticated search syntax, which enabled us to design relatively precise searches. The review of contemporary Polish language through Poliqarp and Monco yielded a database of a few hundred examples containing the expected SVC or PC. The evaluation of these examples -with regard to both their form and meaning - provided the basis for designing searches on the corpora from the earlier epochs.

To search potential cases of SVCs and PCs with wziqc in the Polish of the 19^th and the early 20^th c., we developed two original corpora: a corpus that includes digitalized works of literary texts available in open access within the frame of Wolne Lektury (https://wolnelektury.pl/) and a corpus containing stylistic texts compiled with the search engine FBL Risercz (Gralinski & Liberek 2019), which allowed us to limit the material available on the Internet to the time relevant for our study (i.e. 19^th and early 20^th c.). Both corpora have been annotated grammatically in an automated manner and subsequently searched with the aid of the search engine Korpusomat (https://korpusomat.pl) (Kieras, Kobylinski & Ogrodniczuk 2018). An important advantage of this application is its ability to work on individual sources of texts, the use of an annotation system that is compatible with the system employed in the National Corpus of the Polish Language, and, consequently, a similar syntax of searches, which allows for trans-corpus comparisons of the obtained results. Despite the usefulness of the two abovementioned historical corpora and the results they generated, both of them exhibit important limitations. First, because of the texts used, neither of these corpora complies with representativity requirements (Górski & Lazinski 2012). Second, the set of lexico-grammatical segments that could be searched automatically was limited. Third, due to a lower number of texts included than is the case of NKJP or Monco, the overall number of the collected examples was much lower than is the case of modern Polish corpora discussed in the previous paragraph.

Consequently, while the presence or absence of the wziqc SVC and PC in modern corpora and their frequency constitute significant pieces of information because of the size and representativity of these corpora, in the case of the corpora from the 19^th and early 20^th c., this same presence or absence, and any frequencies, must be treated in terms of probability.

Examples from the 17^th and 18^th c. have been extracted from the Corpus of Polish Texts of the 17^th and 18^th c. with aid of the search engine Korba (https://korba.edu.pl/) (Gruszczynski, Adamiec & Ogrodniczuk 2013). As in the case of the previous analysis, to ensure relative commensuration of the obtained results, we used an annotation system and searching procedures that were compatible with the annotations and searches made in the National Corpus of the Polish Language. In addition to the limitations explained in the paragraph above, which characterize our original 19^th and early 20^th c. corpora, the lexico-grammatical terminology used in the engine Korba (designed to search the 17-18^th c. texts) and the possible query syntax are less sophisticated than is the case of Poliqarp, Monco, and Korpusomat mentioned above.¹⁶

The material allowing for searching for the relevant constructions with wziqc in the 16^th c. was obtained from the Corpus of 16^th c. Old Polish (https://spxvi.edu.pl/korpus/teksty/). This corpus has two important limitations, on top of the limitations exhibited by the historical corpora mentioned thus far. Although the Corpus of 16^th c. Old Polish contains texts that form the basis of the Dictionary of 16^th c. Polish, the corpus itself remains in a preliminary phase of development, which means that the number of available texts is very restricted. Furthermore, because of the stylistic character of the texts included in this corpus, our searches had a lexemebase form instead of grammar-based searches that were typical during our work on corpora mentioned previously. As a result, the selection of examples that were subject to further analysis was semi-automated and semi-manual. That is, from the set of examples generated by automated searches containing a specified segment (i.e., wziqc) rather than a sequence of segments (an SVC, PC, or CC), we selected those in which said segment would constitute part of a larger structure - a SVC, a PC, or a coordinated construction. At this stage, we also performed manual searches in analogue texts (i.e., those that had not been digitalized yet) that were made available to us by the editors of the Dictionary of 16^th c. Polish.

At the end of our empirical research, we conducted searches on texts from 14^th and 15^th c. available in the Digital Corpus of Old Polish Text till 1500 (https://ijp.pan.pl/publikacje-i-materialy/zasoby/korpus-tekstow-staropolskich). This corpus is parsed grammatically, which allowed an automated search with the Korpusomat engine.

Once identified through syntax-driven searches (see above), the bi-verbal sequences with wziqc have been examined as potential representatives of the categories of PC, SVC, and CC (syndetic or asyndetic). The inclusion of an example into the PC and SVC categories has primarily been achieved by means of semantic factors, namely: (a) the (more or less patent) mono-event interpretation of a given sequence and, related to it, (b) the use of the verb wziqc in senses that diverge from its literal meaning 'take'. In such cases, rather than expressing two separate events (simultaneous or consecutive), the bi-verbal structures with wziqc (with or without lexemes that are homophonous with coordinators) express a single (or "compact") event and exhibit a constructional meaning of "emotional emphasis" (Andrason 2018a:607) -i.e., intensity, insistence, immediacy, impatience, irritation, criticism - surprise, spontaneity, and/or effortlessness (Góralczyk 2010; Zinken 2013; Andrason 2018a:607-610). Since our study focuses on formal aspects, we will not provide a detailed semantic analysis of the cited examples.

3.2 Findings

The records from before the 19^th century attest to 32 cases of the use of the verb wziqc in constructions with other verbs: 2 in the 15^th c., 8x in the 16^th c., 17x in the 17^th c., and 5x in the 18^th c. No clear cases of SVCs and PCs - whether canonical or less canonical - are attested. Instead, wziqc appears in a variety of bi-clausal CC constructions.¹⁷

The two cases of bi-verbal structures with wziqc attested in the 15^th c. instantiate a syndetic CC pattern with an overt coordinator i (see 1.a-b below). Orthographic separatrices are not used between the two verbs.¹⁸

In the 16^th c., two main types of constructions with wziqc are attested, all of them CCs, syndetic and asyndetic. Syndetic CC constructions are more common than asyndetic ones. The former are found 6 times, while the latter appear only twice. Two overt coordinators are used, namely, i (see wzyqt[...]y wtozyt 'took [...] and put in' in (2.a)) and a (see wezmie a rozwazy 'will take (in his head) and consider' in (2.b)). The use of i is more common (4x) than that of a (2x). In the two asyndetic examples, the verbal clauses are juxtaposed with no orthographic separatrices such as commas, semi-colons, or hyphens (see wzyqt [...] vkazat 'took (it) showed' in (2.a)).

A similar situation is revealed by texts from the 17^th c. That is, SVCs and PCs with wziqc are unattested. Only CC constructions are found. Among them, syndetic CC constructions with i are the most common, being attested 9 times (see wezmiesz i uwarzysz 'you will take and cook' in (3.a)). In contrast, the coordinator a only features once (see wezmi a odnies 'take and return' in (3.b)). This means that the preference for i rather than a has intensified. This is consistent with the gradual loss of a coordinating function by a and its grammaticalization into an adversative-contrastive connector (see Andrason 2020). Asyndetic examples of coordinated constructions with wziqc are also attested, specifically, in 7 cases (see wziqc przeprosic 'take (as the master) and ask for forgiveness' in (3.c)). As in the 16^th c., two clauses may be juxtaposed without any orthographic sign. However, they may also be separated by a comma or other graphic devices (e.g., <*> or </>).

In the 18^th c., again, only CC constructions with wziqc, both syndetic and asyndetic, are found. Syndetic CC constructions with a are unattested. CC constructions with i are found 3 times (see wziqt i ztupit 'he took and plundered' in (4.a)). Asyndetic CC constructions occur twice. In the two cases attested, a comma is used to separate the juxtaposed clauses (see wezcie, wrzuccie ich 'take and throw them' in (4.b)).

Although, as demonstrated above, no example found up to the 18^th c. can be classified as an SVC or PC, a few uses of the verb wziqc in CC constructions exhibit a certain affinity with SVCs or PCs and may be understood as starting points for the emergence of these two categories in the later centuries. The three most exemplary cases are: wzyqt ie vkazat ludu (2.a), miasto [...] wziqt i ztupit (4.a), and wezcie, wrzuccie ich w ogien (4.b). It is true that, in these examples, wziqc maintains its literal sense 'take' and the two verbs most likely express two independent events. Nevertheless, the pattern [wziqc (i) V2] employed lends itself to certain more compact constructional interpretations. This stems from the following features exhibited by the wziqc CC used in these examples: the TAM inflections of the two verbs coincide; the subject referent is identical; the two verbs share an object which is expressed only once and operates overtly in the entire construction, occupying a construction-initial (4.a), -final (4.b), or -medial position (2.a); even if consecutive, the two events occur immediately one after another and, crucially, form a holistic semantic unit, i.e., presenting a written text, plundering a town, and throwing men into fire - in other words, the verb wziqc contributes rather minimally to the sequential eventhood of the construction, in each case, the main event is the event expressed by V2.²⁰

A similar case, which almost naturally lends itself to further grammaticalization towards a PC and (if the linker is omitted) an SVC, is the earliest example attested in the corpus from the 19^th c. (see 5 below), dated from 1856. In this example, the construction with wziqc exhibits properties that relate it to both CC and PC. On the one hand, the verb wziqc, inflected in the 1^stsg. future wezme, may be understood literally, i.e. 'I will take', since a barrel of honey (miodu beczka) must be taken in one's hands before one may drink from it. Accordingly, the bi-verbal construction is interpreted as the coordination of two clauses that express two consecutive events. On the other hand, as the TAM and subject and object referents of the two verbs coincide and the critical event is not taking the receptacle but drinking from it (wypije 'I will drink'), a more constructional reading, slightly closer to PC is, at least, theoretically possible. This is even more likely because the preposed object miodu beczka, which operates over the entire construction (wezme i wypije), is separated by a comma from the bi-verbal sequence, thus suggesting a more compact interpretation of this collocation.²¹ To conclude, in this example and the three examples discussed in the previous paragraph, albeit not lost, the literal meaning of wziqc is weakened and its relevance to the overall eventhood of the construction is compromised.

The first instances of constructions with wziqc that are not CCs but constitute genuine cases of PC or SVC come from the second half of the 19^th c. To be exact, a text from 1861 attests to an SVC - wezme zaprobuje 'I will (take) try' (see 6 below). In this example, V1 wziqc and the other verb, i.e., V2 zaprobowac 'try' (an equivalent of Modern Polish spróbowac) appear contiguously and are not separated by any orthographic sign, e.g., a comma. Crucially, the verb wziqc does not convey its literal meaning 'take' nor does it introduce an independent event, whether the first in a sequence or simultaneous with another event, contrary to all the examples of wziqc constructions analyzed above. Rather, the 1^st p.sg. future wezme modifies the event expressed by the verb zaprobowuje 'I will try' - it communicates insistence and/or defiance with regard to the intention of writing a comedy (zaprobuje.napisze komedye).

Two other constructional examples that are dated from that same decade are even more critical. In 1867, a metalinguistic note published in the newspaper Kurier Warszawski records two PC constructions with wziqc: wziql i zaspiewal 'he (took and) sang' and wziqlem i siadlem 'I (took and) sat down' (see 6 below). The author of this note comments on a particular usage in contemporary spoken Polish and observes that many people (wiele osób) employ the verb wziqc in front of another verb which expresses the event which they are, were, or will be performing. This suggests that, as is typical of PCs and SVCs, the verb wziqc does not express a separate event but rather modifies the event conveyed by the other verb - the major verb in the construction - in some manner. Indeed, in both uses in (7), the bi-verbal constructions likely express single events related to singing and sitting down, respectively. Accordingly, in those examples, the verb wziqc cannot be interpreted literally as a full verb but rather adds some modal nuances, which due to the lack of context cannot be specified.²² Crucially, the two examples of the mono-event, mono-clausal, and mono-predicative usage of a bi-verbal construction with wziqc that are cited by the author, make use of the linker i and thus attest to a PC. Inversely, a serializing pattern (i.e., an SVC), in which the i linker would be absent, is not mentioned.

Overall, excluding the ambiguous example (6), there are 24 possible cases of SVC and PC constructions with wziqc in the 19^th c. Among all of them, 20 cases (see (7) and (8.a-r) below) exhibit the linker i (lit. 'and') and thus attest to a PC pattern. In most cases, the PC construction used is intransitive, e.g., wziela i zwiedla 'she withered' (8.b), wziela i sfiksowata 'she got crazy' (8.e), wziqt i wyjechat 'he left' (8.f), wziqt ipoptynqt 'he drifted away' (7.g), wezme i uciekne 'I will escape' (8.h); see also wziqtem i siadlem 'I sat down' and wziql i zaspiewal 'he sang' (8.b). Transitive uses with objects operating over the entire wziqc-i PC are also attested, e.g., wezme i nadam konstytucje 'I will give the constitution' (8.c); wziql i zaczql[...] chodzic 'he started visiting' (8.i), wezme i opisze przesladowanie 'I will describe persecutions' (8.n), wziql i wysadzil mnie 'he dropped me off (8.o), and wziely i obdarly [...] liscie 'they stripped off the leaves' (8.p). Sometimes, the transitive object seems to be unexpressed: wzieli i pozwolili 'they allowed (it)' in (8.j) and wziela ipodrzucila 'she flipped (it)' (8.k). The use of reflexive verbs as V2 is also attested, e.g., wziql i ozenil sie 'he got married' (8.a, 8.l, 8.s), wezme ipowiesze sie 'I will hang myself (8.d), wzielaizamknelasie 'she closed herself (8.m), and sie wziql i zlitowal 'he had mercy' (8.r). In this last example, the reflexive pronoun sie, which is originally provided by the argument structure of V2, i.e., zlitowac sie 'have mercy' appears in front of V1 wziqc. This overtly indicates that the bi-verbal sequence is understood in a mono-event, mono-predicative, and mono-clausal manner (see Andrason 2020). It should also be noted that the 19^th c. PCs with wziqc attest to a variety of subjects (i.e., 1^st (e.g., 8.c) and 3^rd (e.g., 8.a); singular masculine (e.g., 8.a) and singular feminine (e.g., 8.b); human-masculine plural (8.j) and non-human-masculine plural (8.p)) and TAM categories (i.e., future (e.g., 8.b) and past (e.g., 8.a)).

In contrast, corpora from the 19^th c. only attest to 4 examples (see 6 and 9.a-c) of SVCs built around the minor verb wziqc, in which the linker i is absent. One of them is inherently intransitive: wziety, zawrócify 'they turned back' (9.a). One makes use of the ethical dative: wzieta, poszta sobie 'she left' (9.b). One exhibits an overt transitive object: wziql, sam zabit [...] kurcze in (9.c). And in one, the object is unexpressed: wezme zaprobuje 'I will try' (6). Out of the four examples, three wziqc SVCs appear in sequences consisting of three (9.a-b) or four (9.c) verbs (including the verb wziqc). It should also be noted that in these three examples (9.a-c), V1 and V2 are separated by a comma. This may suggest a lesser extent of phonological contiguity - perhaps a slight pause between V1 and V2 - and thus a slightly less canonical profile of these SVCs. In contrast, in (6) analyzed previously, no orthographic separatrix is present. As the PC constructions discussed in the paragraph above, SVCs with wziqc are compatible with various types of subjects (i.e., 1^st (see 6) and 3^rd (9.a-c) singular; masculine (9.c) and feminine (9.b) singular; non-human-masculine plural (9.a)) and TAM categories (i.e., future (6) and past (9.a.-c)).

The corpus from the pre-war 20^th c. attests to 14 instances of mono-event, mono-predicative, and mono-clausal bi-verbal patterns with wziqc. Again, the majority of such constructions are PCs with the linker i (12x). These constructions may be intransitive: wziql i umarl 'he died' (9.f), wezme ipojade 'I will go' (10.h), and weziprzeczytaj 'read!' (10.b), or - more commonly - transitive: wezmie i wykrzywi [...] zelazo 'he will bend iron' (10.a), wziela i wygonila [...] panny 'she chased away the girls' (10.c), wezmie i coszepsuje 'she will break down something' (10.d), wezme i wysle cie 'I will send you' (10.g), wezme i sprzedam [...] butelke 'I will sell the bottle' (10.i), wezme i [...] zarzne czlowieka 'I will slaughter the man' (10 .j), and wezme i obtluke [...] reke 'I will break the hand' (10.l). In one case, a predicative complement is used: wezme i zostane szoferem 'I will become a driver' (10.e). In another single case, the wziqc-i PC construction is reflexive: sie wziqc i powiesic 'to kill oneself (10.k). As in (8.r), the reflexive marker sie occupies a construction-initial position, i.e., in front of V1 wziqc rather than V2 powiesic by the valence of which it is required (cf. powiesic sie 'hang oneself). In the 20^th c., we find the first negative use of the wziqc-i PC. Significantly, the negator is used twice, with each verb separately, i.e., nie wezme i nie zarzne I will not slaughter' (lit. I will not take, and I will not slaughter') (10 j). This suggests a less advanced stage of the grammaticalization of a negative bi-verbal sequence into a genuine SVC, and thus its less canonical profile. Similar to the examples from the 19^th c., a variety of subject types (i.e., 1^st (e.g., 10.e), 2^nd (10.b), and 3^rd (e.g., 10.c); masculine (10.f) and feminine (10.c) singular), and TAM categories (i.e., past (e.g., 10.c), future (e.g., 10.e), imperative (10.b), and infinitive (10.k)) are attested.

While a PC with wziqc is common in the pre-war period, there are only two cases in which mono-event, mono-predicative, and mono-clausal constructions with wziqc do not make use of the linker i and instantiate an SVC. One of these examples involves the ethical dative: wez, wypij sobie 'drink!' (11.a). The other involves a transitive object: wez, przeczytaj ten papier 'read this document' (11 .b). In both cases, a comma separates the minor verb wziqc from the major verbs wypic 'drink' and przeczytac 'read'. As previously, this suggests a phonological detachment and thus a less canonical profile of SVCs. Contrary to the variation of subjects and TAM categories used in PC constructions with wziqc from this period, wziqc SVCs only attest to 2^nd person singular subjects and imperatives. This apparent lack of diversity is most likely due to the overall scarcity of the attested examples (compare with the diversity of subjects and TAM categories observed in the 19^th c. as discussed above).

Overall, since the appearance of PCs and SVCs with wziqc in the sixties of the 19^th c. until the end of the New Polish period in 1939, PC patterns are significantly more common than SVC ones. Similarly, the wziqc-i PC attests to a more diverse repertoire of uses: it allows for the placement of the reflexive marker sie in a construction-initial position, i.e., before V1, although it is originally projected by the valency of V2 and is compatible with negative polarity. The wziqc SVC does not attest to these uses. Moreover, the serializing pattern with wziqc tends to separate V1 from V2 by a comma suggesting some degree of phonological detachment and, at least in the 19^th c., usually appears in sequences of three verbs the last of which is headed by the true coordinator i. Before the second half of the 19^th c., only CCs with wziqc are attested. Significantly, before the emergence of the PC and SVC wziqc patterns, syndetic CCs with wziqc as V1, formally more akin to PC, are also more common than asyndetic ones. In syndetic patterns, the presence of the linker i is more frequent than that of a, the use of which ceases in the 17^th c. The entirety of the quantitative evidence resulting from our empirical study is tabulated below:

 

Table 1

 

4. Discussion

In light of the evidence presented in section 3, we can now answer our research question. The SVC with wziqc most probably results from a clause-fusion mechanism, which, as explained in section 2, is one of the three grammatical pathways towards verbal serialization (cf. Aikhenvald 2018). To be exact, mono-event, mono-clausal, and mono-predicative bi-verbal structures built around the minor and first verb wziqc seem to have emerged from original bi-clausal CC structures in which the first conjunct clause contained the full verb wziqc while the latter contained another full verb inflected in the same TAM, person, and number. Inversely, our research does not provide evidence that two other possible sources of SVCs, namely verbal modification or concurrent grammaticalization (Aikhenvald 2018), may have contributed to the formation of the wziqc SVC.

Out of the two hypothetical scenarios possible in case of clause-fusion mechanisms - i.e., from syndetic CC with the coordinator i to a wziqc SVC via a wziqc-i PC or from asyndetic CC directly to a wziqc SVC - the evolution through a PC stage seems to have constituted the principal and stronger drift. This means that syndetic CC is a more likely source of wziqc SVCs than asyndetic CC and/or that the stabilization of wziqc SVCs in Polish is mainly due to the grammaticalization of a syndetic CC pattern and its gradual serialization via a PC stage. This may be postulated given the two clusters of reasons: one cluster concerns the properties of the wziqc-i PC and wziqc SVC in the 19^th and 20^th c., whereas the other cluster concerns the properties exhibited by CCs containing the verb wziqc at earlier developmental stages.

With regard to the properties exhibited by the wziqc-i PC and wziqc SVC the following should be noted: The grammaticalization of the wziqc-i PC advanced faster than was the case of the wziqc SVC. This is visible through dissimilar quantitative and qualitative properties of these two constructions. Quantitatively, both in the 19^th and the first four decades of the 20^th c., wziqc-i PCs are roughly five times more frequent than wziqc SVCs. Qualitatively, the wziqc-i PC exhibits stronger constructional properties than the wziqc SVC, i.e., it allows for more diverse and more "compact" patterns such as the construction-initial placement of the reflexive pronoun sie (attested in the 19^th c.) and negative polarity (attested in the 20^th c.). In contrast, these two types of uses are unattested by the wziqc SVC, which furthermore tends to exhibit a comma between the verb wziqc and V2, suggesting some degree of phonological detachment and thus less advanced or pre-canonical profile. Additionally, most instances of wziqc SVCs attested in the 19^th c. corpora appear in cases where the wziqc + V2 sequence is followed by a third verb headed by a coordinator i. Accordingly, those wziqc SVCs have most likely emerged to the loss of the lexeme i that was placed between V1 wziqc and V2 that was eliminated due to the presence of another i linking V3 to the V1-V2 sequence. That is, in Polish, sequences V1 i V2 i V3 tend to be replaced by sequences V1 (,) V2 i V3. As a result, these SVCs with wziqc would have derived from wziqc-i PCs. Lastly and most critically, the note from 1867 that overtly recognizes the constructional and non-coordinated use of wziqc in bi-verbal but monoevent constructions, only makes reference to the wziqc-i PC. In other words, by the mid-19^thc., wziqc-i PC had already been profoundly grammaticalized in spoken language. Linker-free wziqc SVCs are not mentioned and their exact grammaticalization stage in the spoken language of that period is thus unknown.²⁴

With regard to the properties exhibited by CCs with wziqc in the pre-19^th c. corpora, syndetic CC patterns seem to have been more propitious for grammaticalization into mono-event, mono-predicative, and mono-clausal structures than asyndetic CC patterns. Again, this is visible in dissimilar quantitative and qualitative profiles exhibited by these two types of CCs. Syndetic CCs, in particular those with the coordinator i, are more frequent than asyndetic ones in the 15^th, 16^th, 17^th, and 18^th c. Furthermore, among all CCs that exhibit a more advanced profile whereby the event conveyed by V2 is much more critical for the entire situation than the event conveyed by V1 wziqc - and thus lend themselves to extensions towards mono-eventhood, mono-predicativity, and mono-clausality - syndetic patterns with i are more common than asyndetic ones.

The argument above, which demonstrates the faster grammaticalization of the wziqc-i PC than that of the wziqc SVC, may be complemented by two further synchronic and diachronic observations. In colloquial Polish, CCs have apparently always been predominantly formed with an overt coordinator, especially i, rather than in an asyndetic manner through juxtaposition. As grammaticalization processes tend to originate in frequent patterns -frequency constituting both their cause and effect - it is more likely that the grammaticalization of bi-event, bi-predicative and bi-clausal constructions with wziqc into more compact, monoevent, mono-predicative, and mono-clausal constructions originated in more common patterns, i.e., syndetic CC with i. Furthermore, the lexeme i has always been highly polysemous in Polish and commonly participated in various grammaticalization processes that have led to the development of a wide range of constructions (for the analysis of the various uses of i in Polish, synchronically and/or diachronically, see Patejuk & Przepiórkowski 2012; Przepiórkowski & Patejuk 2014; Patejuk 2015; Andrason 2016b; 2017). Its presence in a putative wziqc i CC pattern would have created further favorable ground for grammaticalization because the construction would contain two constant elements (i.e., wziqc and i) rather than a single one (i.e., the verb wziqc).²⁵ Across languages, more robust patterns are more (easily) grammaticalizable than those that are less robust.

If our conclusion is correct and the wziqc-i PC was grammaticalized first and more intensively than the wziqc SVC, the highly common use of linker-free SVCs in the second half of the 20^thc. and in the 21^st c. - to the extent that it is equally or more common than the wziqc-i PC (Andrason 2018a; Nau et al. 2019) - would be a subsequent development. This increase in the grammaticalization of the wziqc SVC would likely have resulted from the elimination of i - a process which itself may have occurred due to two reasons. First, it may have resulted from morphological or phonological reduction typical of more advanced stages of grammaticalization, in our case, of the wziqc-i PC. This is consistent with the fact that in modern Polish, the element i in wziqc-i PCs is typically phonologically more reduced and weaker than the genuine coordinator i (Andrason 2019a). Second, this gradual weakening and ultimate elimination of i must have increased the overall number of linker-free bi-verbal mono-event, mono-predicative, and mono-clausal constructions with wziqc. Accordingly, by analogy, at least some, initially less common, wziqc SVCs may have replaced the "catastrophically" more common wziqc-i PCs.

Overall, even though our study suggests the evolution from the wziqc-i CC to the wziqc SVC via the wziqc-i PC, the actual development has most likely been messier. That is, although wziqc-i PC is a faster emerging construction than the wziqc SVC, and the change from wziqc-i CC (i.e., syndetic CC with wziqc and i) to wziqc-i PC has constituted a stronger evolutionary drift than the grammaticalization of the wziqc SVC from asyndetic CC with wziqc - which points to an overall wziqc-i CC - wziqc-i PC - wziqc SVC evolutionary sequence - this development has also (probably from its very origin) been accompanied by the development of wziqc CC (i.e., asyndetic CC with wziqc) - wziqc SVC. Although slower and less potent, this drift may have played some role in the loss of the linker and the evolution of the wziqc-i PC into the wziqc SVC. Thus, both types of clause-fusion have most likely operated during the entire grammaticalization process of the wziqc SVC. Such messiness is typical of grammaticalization phenomena in general, especially if a single target construction has evolved from two similar inputs, as is the case of the wziqc SVC.

In addition to answering the research question, the provided evidence allows us to formulate certain important generalizations. First, our study suggests that the first attested step in the development of CC constructions with wziqc - whether syndetic or asyndetic - may have involved the emergence of a semantic imbalance between the events expressed by V1 (wziqc) and V2. While, in a canonical CC pattern, the event of each clause is equally critical for a particular situation (e.g., Jem i czytam 'I eat and read') in less canonical CC patterns, the key event is conveyed by V2, while the relevance of the event expressed by wziqc is secondary. Furthermore, according to our data, such input CC constructions were transitive with a shared object that was expressed only once, either before V1, after V2, or between the two verbs, and seems to belong to both verbs simultaneously. Second, the historical evidence presented allows us to propose an intermediary stage between asyndetic CC and the SVC, namely that of an empty or pseudo-comma. That is, similar to the empty linker or pseudo-coordinator - which is formally related to the coordinator from which it derives - an empty or pseudo-comma is a phonological comma rather than a syntactic one, even though "deriving" from the former. Accordingly, instead of separating verbal clauses as is typical of asyndetic CC (e.g., Je, pije 'He eats, drinks'), this type of comma does not mark new clauses but indicates that the two verbs, now forming a single clause, are phonologically separated by a pause of, at least, minimal duration.

The fact that the first instances of the serializing patterns with wziqc - both PC and SVC - are attested in the 19^th c. opens a possibility of language-external influence. Since the late 18^th c. (to be exact, in three stages, i.e., in 1772, 1793, and 1795) till the end of World War I in 1918, the Polish territory was partitioned by Prussia, the Austrian(-Hungarian) Empire, and Russia. Crucially, in the 19^th c., the PC construction with the verb vzjat' - a cognate of wziqc - was quite widespread in Russian, as is evidenced by the Russian National Corpus (www.ruscorpora.ru). In contrast, German - whether in Prussia or Austria - did not include an equivalent construction in its verbal repertory. Interestingly, in 45,5% of our examples, the influence of Russian is highly possible: the author lived in the Russian partition of Poland or the source (i.e., a particular journal) was published in the territory incorporated into the Russian Empire. In 42,5%, such influence cannot be ruled out, although it cannot clearly be demonstrated either. Nevertheless, in 12%, Russian influence is virtually impossible: the source of these examples was published in the Austrian or Prussian partition. Given this last observation, as well as given that the use of 'take'-verbs in PCs and SVCs is typologically highly common - being widely attested in Indo-European languages of Europe (e.g., English, Swedish, Norwegian, Danish, Spanish, Italian, Portuguese, Romanian, Greek, Albanian, Lithuanian, Latvian, Bulgarian, Serbo-Croatian) and languages from other phyla (also those spoken in Europe, e.g. Hungarian, Estonian, Finnish, and Sami; see Coseriu 1966; Aikhenvald 2006; Zinken 2013; Nau et al. 2019) - the emergence and grammaticalization of the PC and SVC with wziqc may be to (a considerable extent) a language-internal phenomenon. The presence of similar constructions in Russian may, however, have created a favorable ground for this process. This view concords with the conclusion reached by Nau et al. (2019:290) according to whom "the constructions [with 'take' in Baltic and Slavic] have been developing [...] mostly language-internally, and that similarities across languages are connected to general factors rather than to [...] mutual influences." Nevertheless (pace Weiss 2012), the areal properties of bi-verbal 'take' constructions in Eastern Europe (Polish, Russian, Ukrainian, Lithuanian, Latvian, and Estonian) and Scandinavia (Finnish, Swedish, Norwegian, and Danish) should not be underestimated.

Although we have fulfilled our research plan - we have answered our research question and determined the most likely origin of the wziqc SVC in Polish - our conclusions may not be taken incautiously. Their limitations must be acknowledged. First, as has already been mentioned in section 3, due to the scarcity of the available material and examples, the generalizations on which these data draw and thus our conclusions can only be approximate. (For instance, while our evidence demonstrates that the wziqc PC has been fully grammaticalized in spoken Polish since at least the mid-19^th c., the actual time of its grammaticalization may have been earlier.) Second, colloquial Polish, and thus the variety of Polish in which the grammaticalization of the wziqc-i PC and wziqc SVC must have occurred first - indeed, both constructions are still more typical features of spoken and dialectal Polish than the standard written language even though they appear in some literature - is poorly represented in the historical corpora analyzed. This means that the possibility of encountering the relevant PC and serialized constructions is lower and does not, most likely, correspond with what typified the spoken Polish language of these periods. Third, in texts from pre-19^th c. corpora, punctuation is variable (sometimes even erratic) as the orthographic rules were still being crystalized. Thus, the relevance of punctuation for the inclusion (or exclusion) of the examples dated from those periods into the CC, PC, and SVC categories should be considered with much caution.

In light of the above limitations - especially the second one - a new avenue of further research may be proposed, namely, a study of court oaths (roty sqdowe) recorded since the 14^th c. This type of historical material, which reflects more closely the spoken Polish language and its dialectal usage, may provide a unique opportunity to detect earlier cases of the wziqc-i PC and wziqc SVC than those revealed in the present study. Indeed, if any texts, these texts should attest to the first instances of both constructions with wziqc in Polish. As the online corpus of such court oaths has recently been made available (https://rotha.ehum.psnc.pl/), we plan to embark on this research venture in the near future.

 

5. Conclusion

The present article studied the emergence of an SVC with wziqc in Polish by making use of a dynamic prototype-driven approach to linguistic categorization and by reviewing the historical corpora from the first texts available in the14^th c. until the end of the New Polish period in 1939. The evidence suggests that the wziqc SVC has resulted from clause-fusion mechanisms of which the evolution from the syndetic CC with the coordinator i to a wziqc SVC via a wziqc-i PC seems to have constituted the principal and stronger drift than the more direct evolution originating in the asyndetic CC with wziqc. Although our research points to syndetic CC patterns with i as the main source of the wziac SVC (through the stage of the wziqc-i PC), asyndetic CC patterns may have also played some role in this development. Overall, the evolution of the wziqc SVCs and its profound grammaticalization in modern Polish has, probably from its incipient stages, been a messy bi-source phenomenon.

 

Abbreviations

CC - conjunctive coordination; EDAT - ethical dative pronoun (sobie); F - feminine; FUT -future tense; HM - human-masculine; IMP - imperative; INF - infinitive; M - masculine; NHM - non-human-masculine; PAST - past tense; PC - pseudo-coordination; PL - plural; REFL - reflexive pronoun (sie); SG - singular; SVC - serial verb constructions; TAM - tense, aspect, mood.

 

References

Aikhenvald, A. 2006. Serial verb constructions in typological perspective. In A. Aikhenvald and R. M. W. Dixon (eds.) Serial Verb Constructions: A Cross-linguistic Typology. Oxford: Oxford University Press. pp. 1-68. https://doi.org/10.1093/oso/9780198791263.003.0010        [ Links ]

Aikhenvald, A. 2011. Multi-verb constructions: Setting the scene. In A. Aikhenvald and P. Muysken (eds.) Multi-verb Constructions: A View from the Americas. Leiden: Brill. pp. 1-26. https://doi.org/10.1163/ei.9789004194526.i-313.8'        [ Links ]

Aikhenvald, A. 2018. Serial Verbs. Oxford: Oxford University Press.         [ Links ]

Andrason, A. 2016a. From vectors to waves and streams: An alternative approach to semantic maps. Stellenbosch Papers in Linguistics 45: 1-29. https://doi.org/10.5774/45-0-211        [ Links ]

Andrason, A. 2016b. The coordinators i and z in Polish: A cognitive-typological approach, part 1. Lingua Posnaniensis 58(1): 7-24. https://doi.org/10.1515/linpo-2016-0001        [ Links ]

Andrason, A. 2017. The coordinators i and z in Polish: A cognitive-typological approach, part 2. Lingua Posnaniensis 59(2): 7-22. https://doi.org/10.1515/linpo-2017-0009        [ Links ]

Andrason, A. 2018a. The WZIAC gram in Polish. A serial verb construction, or not? STUF -Language Typology and Universals 71(4): 577-629. https://doi.org/10.1515/stuf-2018-0022        [ Links ]

Andrason, A. 2018b. From coordination to verbal serialization - The pójsc (serial verbal) construction in Polish. Research in Language 16(1): 19-46. https://doi.org/10.2478/rela-2018-0001        [ Links ]

Andrason, A. 2019a. A pseudo-coordinated Serial Verb Construction "wziqc i V2" in Polish. Slovo a Slovesnost 80: 163-191.         [ Links ]

Andrason, A. 2019b. The conjunctive coordinator NA in Xhosa - Its categorial status and map of polyfunctionality. Italian Journal of Linguistics 31(1): 3-60.         [ Links ]

Baba, S. and S. Mikolajczyk. 1974. O orzeczeniu szeregowym. Poradnik Jezykowy 4: 117-128.         [ Links ]

Banko, M. 2012. Wez przestan! https://sjp.pwn.pl/poradnia/haslo/Wez-przestan;13293.html [accessed 08.04.2017].

Bartminski, J. 1978. Swoiste formy orzeczen w jezyku ustnym. In S. Grabias, J. Mazur and K. Pisarkowa (eds.) Studia nad skladniq polszczyzny mówionej. Wroclaw: Zaklad Narodowy im. Ossolinskich. pp. 159-175.         [ Links ]

Bisang, W. 2009. Serial verb constructions. Language and Linguistics Compass 3(3): 792-814. https://doi.org/10.1111/i.1749-818x.2009.00128.x        [ Links ]

Bowern, C. 2008. The Diachrony of Complex Predicates. Diachronica 25(2): 161-185. https://doi.org/10.1075/dia.25.2.03bow        [ Links ]

Coseriu, E. 1966. 'Tomo i me voy'. Ein Problem vergleichender europäischer Syntax. Vox Romanica 25: 13-55.         [ Links ]

Croft, W. 2003. Typology and Universals. Cambridge: Cambridge University Press. https://doi.org/10.1017/cbo9780511840579        [ Links ]

De Vos, M. 2004. Pseudo coordination is not subordination. In L. Cornips and J. Doetjes (eds.) Linguistics in the Netherlands 2004. Amsterdam: John Benjamins. pp. 181192. https://doi.org/10.1075/avt.21.20vos        [ Links ]

De Vos, M. 2005. The Syntax of Pseudo-coordination in English and Afrikaans. Utrecht: LOT.         [ Links ]

Dixon, R. M. W. 2006. Serial verb constructions: Conspectus and coda. In A. Aikhenvald and R. M. W. Dixon (eds.) Serial Verb Constructions: A Cross-linguistic Typology. Oxford: Oxford University Press. pp. 338-350. https://doi.org/10.1515/lingty-2017-0008        [ Links ]

Fortuin, E. L. 2000. Polysemy or Monosemy: Interpretation of the Imperative and the Dative-Infinitive Construction in Russian. Amsterdam: Institute for Logic, Language and Computation.         [ Links ]

Gebka-Wolak, M. 2011. Pozycje skladniowe frazy bezokolicznikowej we wspólczesnym zdaniu polskim. Torun: Wydawnictwo Uniwersytetu Mikolaja Kopernika.         [ Links ]

Gebka-Wolak, M. 2012. Wez sie zastanów! Problem status gramatycznego przyimperatywnego wez. Poradnik Jezykowy 4: 49-63.         [ Links ]

Gebka-Wolak, M. and A. Moroz. 2017. Idz zadzwon, idz sie wypchaj - dwurozkaznikowe struktury we wspólczesnej polszczyznie potocznej. Paper presented at the Conference 'Jezyki Slowianskie w Badaniach Frazeologicznych', November 9, 2017, Torun, Poland.

Gebka-Wolak M. and A. Moroz. 2021. Serial verb constructions containing the element idz in contemporary colloquial Polish. Slovo a Slovesnost 82: 209-231.         [ Links ]

Giusti, G., V. Di Caro and D. Ross. 2022. Pseudo-coordination and multiple agreement constructions. A overview. In G. Giusti, V. Di Caro and D. Ross (eds.) Pseudo-Coordination and Multiple Agreement Constructions. Amsterdam: John Benjamins. pp. 1-32.         [ Links ]

Góralczyk, I. 2010. "Wez, pokochaj smoka". The wziqc construction in Polish: A Construction Grammar analysis. Prace Jezykoznawcze 12: 89-104.         [ Links ]

Górski R. L. and Lazinski M. 2012. Reprezentatywnosc i zrównowazenie korpusu. In A. Przepiórkowski, M. Banko, R. L. Górski, and B. Lewandowska-Tomaszczyk (eds.) Narodowy Korpus JezykaPolskiego. Warszawa: Wydawnictwo Naukowe PWN. pp. 25-36.         [ Links ]

Górski R.L., M. Król and M. Eder. 2019. Zmiana w jezyku. Studia kwantytatywno-korpusowe. Kraków: Instytut Jezyka Polskiego PAN.         [ Links ]

Gralinski F. and J. Liberek. 2019. Przydatnosc tradycyjnych opisów leksykograficznych w dowodach z opinii bieglego jezykoznawcy dotyczacych zniewazenia. Na przykladzie rzeczownika pedal i w kontekscie mozliwosci programu do automatycznego wyszukiwania danych FBL Risercz. In M. Zasko-Zielinska and K. Kredens (eds.) Lingwistyka kryminalistyczna: teoria i praktyka. Wroclaw: Uniwersytet Wroclawski: Quaestio. pp. 343376.         [ Links ]

Gruszczynski W., D. Adamiec and M. Ogrodniczuk. 2013. Elektroniczny korpus tekstów polskich z XVII i XVIII w. (do 1772 r.). Polonica 33: 311-318. https://doi.org/10.33896/porj.2020.8.3        [ Links ]

Haspelmath, M. 2003. The geometry of grammatical meaning semantic maps and cross-linguistic comparison. In M. Tomasello (ed.) The New Psychology of Language. Mahwah: Lawrence Erlbaum Associates. pp. 211-242. https://doi.org/10.4324/9781410606921-11        [ Links ]

Haspelmath, M. 2004. Coordinating constructions: An overview. In M. Haspelmath (ed.) Coordinating Construction. Amsterdam: John Benjamins. pp. 1-40. https://doi.org/10.1075/tsl.58.03has        [ Links ]

Haspelmath, M. 2007. Coordination. In T. Shopen (ed.) Language Typology and Syntactic Description, Vol. 2: Complex Constructions. Cambridge: Cambridge University Press. pp. 151. https://doi.org/10.1017/cbo9780511619434.001        [ Links ]

Hilpert, M. and C. Koops. 2008. A quantitative approach to the development of complex predicates. The case of Swedish Pseudo-Coordination with sitta "sit". Diachronica 25(2): 240259. https://doi.org/10.1075/dia.25.2.06hil        [ Links ]

Hoeksema, J. and D. Napoli. 1993. Paratactic and Subordinative So. Journal of Linguistics 29(2): 291-314. https://doi.org/10.1017/s0022226700000347        [ Links ]

Hopper, P. and E. Traugott. 2003. Grammaticalization. Cambridge: Cambridge University Press. https://doi.org/10.1017/cbo9781139165525        [ Links ]

Janda, L. 2015. Cognitive linguistics in the year 2015. Cognitive Semantics 1: 131-154. https://doi.org/10.1163/23526416-00101005        [ Links ]

Johannessen, J. B. 1998. Coordination. Oxford: Oxford University Press.         [ Links ]

Kieras W., L. Kobylinski and M. Ogrodniczuk. 2018. Korpusomat - A tool for creating searchable morphosyntactically tagged corpora. Computational Methods in Science and Technology 24(1): 21-27. https://doi.org/10.12921/cmst.2018.0000005        [ Links ]

Kleszczowa K. 1991. Historyczna motywacja refleksywnych postaci czasowników. Studia Polonistyczne 19: 91-99.         [ Links ]

Komorowska, E. 2008. Pragmatyka dyrektywnych aktów w jezyku polskim. Szczecin & Rostock: Print Group.         [ Links ]

Kor Chahine, I. 2007. O vozmozhnom puti gramatikalizacii russkogo vzjat'. Russian Linguistics 31: 231-248.         [ Links ]

Królak, E. and K. Rudnicka. 2006. Selected aspects of directives in Polish. Revista Espanola de Lingüistica Aplicada 19: 129-142.         [ Links ]

Muysken, P. and T. Veenstra. 1994. Serial verbs. In J. Arends, P. Muysken and N. Smith (eds.) Pidgins and Creoles: An Introduction. Amsterdam: John Benjamins. pp. 289-301. https://doi.org/10.1075/cll.15.30muy        [ Links ]

Nau, N., K. Kozhanov, L. Lindström, A. Laugalienè and P. Brudzynski. 2019. Pseudocoordination with 'take' in Baltic and its neighbours. Baltic Linguistics 10: 237-306. https://doi.org/10.32798/bl.365        [ Links ]

Palancar, E. 2012. Clausal juxtaposition and subordination: New light from Otomi. In V. Gast and H. Diessel (eds.) Clause Combining in Cross-linguistic perspective. Berlin: De Gruyter Mouton. pp. 37-76. https://doi.org/10.1515/9783110280692.37        [ Links ]

Patejuk, A. 2015a. Unlike Coordination in Polish: An LFG Account. PhD Dissertation, Polish Academy of Sciences.         [ Links ]

Patejuk, A. and A. Przepiórkowski. 2012. Lexico-semantic coordination in Polish. In M. Buttand T. King (eds.) Proceedings of the LFG12 Conference. Stanford: CSLI Publications. pp. 461-478.         [ Links ]

Perczynska, N. 1975. Wybrane cechy sktadniowo-stylistyczne polszczyzny mówionej. Wroclaw: Zaklad Narodowy im. Ossolinskich.         [ Links ]

Pezik, P. 2020. Budowa i zastosowania korpusu monitorujacego MoncoPL. Forum Lingwistyczne 7: 133-150. https://doi.org/10.31261/fl.2020.07.11        [ Links ]

Przepiórkowski, A., M. Banko, R. L. Górski and B. Lewandowska-Tomaszczyk (eds.). 2012. Narodowy Korpus Jezyka Polskiego. Warszawa: Wydawnictwo Naukowe PWN.         [ Links ]

Przepiórkowski, A. and A. Patejuk. 2014. Koordynacja leksykalno-semantyczna w systemie wspólczesnej polszczyzny (na materiale Narodowego Korpusu Jezyka Polskiego). Jezyk Polski 94(2): 104-115.         [ Links ]

Sledz, A. 2001. O pewnej specyficznej funkcji slów 'wez' i 'wezcie' w mowie potocznej. Poradnik Jezykowy 7: 18-22.         [ Links ]

Visser, F. 1969. An Historical Syntax of the English Language. Part Three: Syntactical Units with Two Verbs. Leiden: Brill.         [ Links ]

Weiss, D. 2000. Russkie dvojnye glagoly: kto xozjain, a kto sluga? In L.L. Iomdin and L.P. Krysin (eds.) Slovo v tekste i v slovare. Sbornik statej k semidesjatiletiju akademika Ju.D. Apresjanu. Moskva: Jazyki russkoj kultury. pp. 356-378.         [ Links ]

Weiss, D. 2003. Russkie dvojnye glagoly i ix sootvetstvija v finnougorskix jazykax. Russkij jazyk v naucnom osvescenii 2(6): 37-59.         [ Links ]

Weiss, D. 2007. The grammar of surprise: The Russian construction of the type Koska vzjala da umerla ' Suddenly, the cat died'. In T. Reuther, L. Wanner and K. Gerdes (eds.) Proceedings of the 3rd International Conference on Meaning - Text - Theory. München: Otto Sagner. pp. 427-436.         [ Links ]

Weiss, D. 2008. Vremennaja sootnesennost' dvojnych glagolov soversennogo vida. In G. Kustova (ed.) Dinamiceskie modeli: Slovo. Predlozenie. Tekst. Sbornik statej v test' E. V. Paduchevo. Moskva: Jazyki slavjanskix kul'tur. pp. 155-177.         [ Links ]

Weiss, D. 2012. Verb serialization in North East Europe: The case of Russian and its Finno-Ugric neighbours. In B. Wiemer, B. Wälchli and B. Hansen (eds.) Grammatical Replication and Grammatical Borrowing in Language Contact. Berlin, Boston: De Gruyter Mouton. pp. 611-646. https://doi.org/10.1515/9783110271973.611        [ Links ]

Weiss, D. 2013a. Dvojnye glagoly v russkoj razgovornoj reci v zerkale Nacionalnogo korpusa russkogo jazyka: formy imperativa mnozestvennogo cisla. In E. Velmezova (ed.) Contributions suisses au XVe congres mondiales des slavistes ά Minsk. Bern: Peter Lang. pp. 319-341.         [ Links ]

Weiss, D. 2013b. Russian double verbs in the 1st pl Imperative. Wiener Slawistischer Almanach 85: 165-175.         [ Links ]

Zinken, J. 2013. Reanimating responsibility: The wez-V2 (take-V2) double imperative in Polish interaction. In Thielemann and P. Kosta (eds.) Approaches to Slavic Interaction. Amsterdam: John Benjamins. pp. 35-61. https://doi.org/10.1075/ds.20.04zin        [ Links ]

 

 

1 However, the research on the bi-verbal constructions with the verb wziqc (including the wziqc SVC), especially with regard to the semantic-pragmatic characteristics of their imperative variants, has a long tradition in Polish, as illustrated by the following works presented by Baba & Mikolajczyk (1974), Perczynska (1975), Bartminski (1978), Sledz (2001), Królak & Rudnicka (2006), Komorowska (2008), Góralczyk (2010), Gebka-Wolak (2012), Banko (2012), and Zinken (2013). For a more detailed review of these studies and their respective analyses of the patterns built around the verb wziqc, consult Andrason (2018a; 2019a). Concerning the equivalents of the wziqc PC and SVC, in other Slavonic languages, especially Russian, i.e., the so-called "vzjat' constructions" (Weiss 2007), consult Fortuin (2000), Kor Chahine (2007), Weiss (2003; 2007; 2008; 2012), and Nau et al. (2019). Regarding serial and/or double verb constructions in Russian more generally, see also Weiss (2000; 2012; 2013a; 2013b).
2 Within the dynamic prototype-approach adopted by Andrason, pseudo-coordination is viewed as a non-canonical SVC (Andrason 2018a; 2019a).
3 Similarities between Polish SVCs built around the verb pójsc and other motion verbs, on the one hand, and juxtaposition and thus asyndetic coordination, on the other hand, are also noted by Gebka-Wolak & Moroz (2021). For more detailed analyses of serializing and/or pseudo-coordinated patterns built around motion verbs in Polish consult Gebka-Wolak (2011:68-71; 2012:61), Gebka-Wolak & Moroz (2017; 2012), Andrason (2018b).
4 The regular exception is a consecutive type of coordination.
5 SVCs may also consist of more than two verbs (Aikhenvald 2006; 2018).
6 The contiguity of the two verbs is not included as a prototypical feature is some studies as it is often violated in SVCs attested across languages.
7 Of course, PC violates many other prototypical features of CC, since it complies with several properties typical of SVCs.
8 The two other possible scenarios leading to SVCs are verbal modification and concurrent grammaticalization (Aikhenvald 2018).
9 This type of development has been attested in English and Romance languages (see Aikhenvald 2018:199 citing Visser 1969:1399).
10 In these two cases, a construction could be viewed as less canonical PC, inversely closer to an SVC, albeit still exhibiting a non-canonical serializing profile.
11 The loss of a pseudo-coordinator implies an intermediate stage of a reduced junction.
12 Juxtaposition may exhibit semantic and syntactic properties more typical of coordination or, inversely, subordination, thus yielding a gradual coordination-subordination scale (see Hoeksema & Napoli 1993; Palancar 2012).
13 A historical corpus is a corpus that represents a certain time in the history of a language; a diachronic corpus is a series of such historical corpora (Górski, Król & Eder 2019:14).
14 These are categories compatible with the wziqc SVC in modern Polish (Andrason 2018a).
15 We have disregarded separations larger than two segments as the probability of encountering relevant constructions was very low.
16 The size of the 17-18* c. corpus is also smaller than the 19-early 20 corpora mentioned above.
17 The adopted methodology of searches has not revealed bi-verbal structures containing wziqc in the corpus from the 14 c. methods. (There may, however, be examples involving bi-verbal structure built around the (finite forms of) wziqc when the separation between the two verbs exceeds two segments.) There are also cases of bi-verbal sequences with wziqc in court oaths ('roty') from the 14 and 15 centuries. Due to technical problems, roty have not been included in our corpus (see Section 4).
18 It should be noted that the punctuation of the texts from the 15, 16, 17 and 18 centuries is relatively unstable. This stems from the lack of orthographic canon and the fact that such orthographic rules were still in the process of development. Therefore, the significance of punctuation for the interpretation of the examples from those periods should be considered with much caution (example (5) is a case in point).
19 In the numbered examples, we will only gloss the two verbs that form the relevant constructions, i.e., CC, PC, and SVC. The spelling used in our examples reflects the spelling found in the corpora from which these examples have been extracted.
20 Therefore, if the three examples are pronounced in a more conjunctive manner - e.g., with no pause or contouring - they could be interpreted by modern Polish speakers as perhaps less canonical types of SVCs or PCs.
21 As mentioned in footnote 18 above, the presence or absence of commas in example 5 should be taken with caution. Indeed, the punctuation of L§towski's tragedy is very odd with commas being used excessively, even in places where they are certainly not required by syntax and/or prosody.
22 The verb wziqc in wziql i zaspiewal cannot be interpreted literally because it is impossible to take something and sing it. Similarly, wziqc in wziqlem i siadlem cannot be read literally because siqsc ' sit down' is an intransitive verb incompatible with any object, which is always implicit in the literal meaning of wziqc 'take'.
23 The grey color indicates that we did not search CCs in 19 and 20 c. as, in these periods, both PCs and SVCs are already attested.
24 It should however be noted that the first SVC with wziqc is attested a few years earlier (1861) than a PC (1867). We do not see these 6 years of difference as significant. This more likely stems from the scarcity of historical material that is available (see the discussion of limitations further below in this section).
25 It should be noted that, at least, since the 16 c., wziqc has been used in a range of senses in constructions with nouns and verbs, e.g., infinitives. This broad semantic potential of the verb wziqc and the compatibility with both literal/concrete and extended/more abstract or grammatical meanings has very probably created a favorable ground for the use of the verb wziqc in constructions that have evolved into the wziqc-i PC and wziqc SVC.