Assessing The Role of Pattern and Matter

Assessing the role of pattern and matter replication in the development of Polish
komentarze metatekstowe based on infinite verba dicendi
Polish employs a whole range of deverbal komentarze metatekstowe (metatextual comments,

henceforth KM), such as ogólnie mówiąc ‗generally speaking‘, szczerze powiedziawszy
‗frankly speaking (lit. having said)‘ or wła ciwie biorąc ‗strictly speaking (lit. taking)‘. Those
based on infinite verba dicendi are special because the majority of them represent a
semiproductive formation pattern (cf. Birzer accepted a). Furthermore, we know from the
history of other Slavonic languages (cf. Birzer 2012 for Russian and Birzer (accepted b) for
Croatian) that structurally similar items in the respective languages developed under influence
of language contact. Therefore, the aims of this paper are, firstly, to explore which role
language contact, and more specifically, pattern and matter replication played in the
development of Polish komentarze metatekstowe based on verba dicendi, and, secondly, to
compare the outcome of the process in Polish with Russian and Croatian.
This paper is organized as follows: In the first section we will describe the morphological and
(morpho)syntactic characteristics that allow to identify the research object. The second
section describes the functioning of the research object in Contemporary Polish, to be
followed in the third section by a description of its historical development and the role of
language contact therein. The fourth, concluding section compares the role of replication
processes in the historical development of Polish deverbal KM with Russian and Croatian.
1. Defining the research object

Polish linguistics offer an abounding amount of literature on metatextual issues with
Wierzbicka (especially her seminal article 1971) and Grochowski (1983; 1986a; 1986b and
1998, to mention just a few) among the most renowned researchers working in this field.
Therefore, it is not too surprising that there circulate several terms which are of relevance for
our research object. Three of them deserve closer inspection: wyrażenie metatekstowe
(‗metatextual expression‘), operator metatekstowy (‗metatextual operator‘) and komentarz
metatekstowy (‗metatextual comment‘).1 Wyrażenie metatekstowe is the superordinate term
that roofs operator and komentarz metatekstowy (this and the following cf. abowska 2009Ś
180-181). Particles and conjunctions form the ―[d]wie główne klasy operatorów
metatekstowych [two main classes of metatextual operators – translation S.B.]‖ ( abowska
1 For reasons of space, we refrain from an all-encompassing critical discussion of terminology, as this is already
provided in Birzer (2015).
1
2009: 180) – although abowska does not elaborate on minor subtypes – whereas the
komentarze metatekstowe ‗metatextual comments‘ form a category of their own. The
delineation is on purely formal groundsŚ metatextual comments ―występują na powierzchni
tekstu w postaci skróconych zdań, najczę ciej zwrotów imiesłowowych, […]ś mogą to być
wyra enia samodzielne, wkomponowane w składnię wypowiedzenia, wyra enia wtrącone,
występujące w konstrukcjach tzw. wyodrębnionych [occur on the text surface in the form of
reduced clauses, most frequently in the form of adverbial participle clauses […]ś they can be
autonomous expressions integrated into the syntax of the utterance or parenthetical
expressions figuring in so-called detached constructions – translation S.B.]‖ ( abowska 2009Ś
181). Given that there is no strict regulation on the position of adverbial participles
(henceforth APs) in Polish (cf. Weiss 1977; Feret 2005) we may assume that the syntactic
status of a parenthesis is a characteristic trait of KMs.
It is not surprising that the overwhelming majority of works (see Czapiga 2006: 7-20 and
Grochowski et al. 2014: 19-26 for accounts of the research history) on metatextual
expressions (henceforth ME) has been concerned with operators. Quite often, this equals the
study of particles, although the extension of particle is rather fuzzy. Grochowski et al. (but
cf., among others, also Grochowski 1986b) illustrate this to the point by stating that their
conception of particle does not correspond to that of (school) grammars (2014: 26-27) or
other linguists (2014: 27-28).2 Research on parentheses has been conducted from a very early
stage on (cf. Grochowski 1983), but with less intensity, probably due to their high variability
in morphosyntax (e.g. they may contain a finite, infinite or no verbal element at all),
semantics as well as number (and partially order) of constituents (for the probably most
comprising and latest description of research on parentheses in general cf. Dehé & Kavalova
2007; for an assessment of parentheses in the light of grammaticalization theory cf.
Kaltenböck, Heine & Kuteva 2011). Among the more recent work on Polish parentheses,
Wołk‘s analysis of powiedzmy ‗let‘s say‘ (2007) should be mentioned. She describes the
possibilities of syntactic positioning (2007: 6) and prosodic features (2007: 7) of the item
under investigation, but, unfortunately, does not elaborate on the question whether different
2 Unfortunately, we must state that the defining traits of particles in Grochowski et al. (2014) are contradicted
by the choice of analyzed items. Among others, constructions based on verba dicendi, the more so if the verbal
form combines with several synonymous adverbs, are denied the status of particle (Grochowski et al. 2014:
29), yet krótko ówią i sho t lit. sho tl speaki g is o e of the a al zed pa ti les G o ho ski et al.
2014: 160-162), although it bases on a verbum dicendi and is paralleld, among others, by synonymous najkrócej
ówią ost sho tl speaki g . Między a i ówią i p i ate speaki g G o ho ski et al. 2014: 299-300)
and nawiasem mówiac the a lit. i a kets speaki g G o ho ski et al. : -304) are also
analyzed despite their verbal basis. Furthermore, tak by rzec so to sa G o ho ski et al. : -141) is an
example for a verb-based item whose synonymity with tak y powiedzie also relies on the verbal basis.
2
syntactic positions influence the readings of powiedzmy, as they do, e.g., in the case of
ogólnie mówiąc (cf. Birzer 2015).
With regard to our research object, O óg (1991) deserves special mention, as he gives a
survey of ME based on inflectional forms – both finite and infinite – of various verba dicendi
and cogitandi in spoken Polish. Although this paper has to be considered seminal both for the
role of ME in spoken Polish and for the usage of verba dicendi therein, it also suffers from
some drawbacks, as we will see in the discussion.
The first critical point concerns the delineation of the research object. This problem is also
mirrored in the usage of terminologyŚ O óg himself calls the items under investigation
jednostki metatekstowe ‗metatextual items‘ (1991Ś 184), elementy metatekstowe (1991: 184)
and operatory [metatekstowe] (1991: 184). On the one hand, the fact that the distinction
between operatory and komentarze metatekstowe was introduced only after the publication of
O óg‘s paper in Wajszczuk (2005)3 certainly explains part of the terminological confusion.
On the other hand, the labels jednostka ‗item‘ and element ‗element‘ imply no size of the unit
under research. This is problematic, as verba dicendi display a rather complex semantic
structure with three semantic participants – emitting agent, addressee and content; syntactic
realization of the latter may be highly complex, ranging from a ―simple‖ NP over a PP to a
complementizer clause. Therefore, valid diagnostic criteria are needed to distinguish verba
dicendi in metatextual function from other occurrences. O óg (1991: 184) offers the so-called
―self-correction test (test korekty)‖ as diagnosticsŚ he argues that ME cannot be subject to self-
correction whereas other constituents of a clause are. This sounds rather convincing if one
considers only the ME wiesz ‗you know‘, słuchaj ‗listen.2.IMP‘, co ci powiem ‗I will tell you
something‘, rozumiesz ‗you understand‘, mówię ci ‗I tell you‘ in O óg‘s example (1). They do
not allow the insertion of the self-correction marker chciałem powiedzieć ‗I wanted to say‘
followed by another metatextual expression (1i -1v), whereas the arguments of ―normal‖ verbs
do (1vi).
(1) wiesz słuchaj co ci powiem /wczoraj Janek rozumiesz przyszedł z kina całkowicie
zdenerwowany mówię ci.
‗you know, listen, I will tell you something / yesterday Janek, you understand, returned
from the cinema completely enervated, I tell you.‘
3 Unfortunately, this work is not available and can be perceived only via abowska (2009) and Grochowski et
al. (2014).
3
(1i) *wiesz chciałem powiedzieć rozumiesz /wczoraj Janek przyszedł z kina całkowicie
zdenerwowany.
‗you know, I wanted to say, you understand / yesterday Janek returned from the cinema
completely enervated.‘
(1ii) *słuchaj chciałem powiedzieć rozumiesz /wczoraj Janek przyszedł z kina całkowicie
zdenerwowany.
‗listen, I wanted to say, you understand / yesterday Janek returned from the cinema
(1iii) *coś ci powiem chciałem powiedzieć słuchaj /wczoraj Janek przyszedł z kina
całkowicie zdenerwowany.
‗I will tell you something, I wanted to say, listen / yesterday Janek returned from the
cinema completely enervated.‘
(1iv) *wczoraj Janek rozumiesz chciałem powiedzieć wiesz przyszedł z kina całkowicie
zdenerwowany.
‗yesterday Janek, you understand, I wanted to say, you know, returned from the cinema
v
(1 ) *wczoraj Janek przyszedł z kina całkowicie zdenerwowany mówię ci chciałem
powiedzieć coś ci powiem.
‗yesterday Janek returned from the cinema completely enervated I tell you, I wanted to
say, I will tell you something.‘
(1vi) wczoraj Janek chciałem powiedzieć Jurek przyszedł z kina całkowicie zdenerwowany.
‗Yesterday Janek, I wanted to say, Jurek Janek returned from the cinema completely
enervated‘
All discussed ME serve the same functionsŚ they may be used either to guide the hearer‘s
attention to the following information, or as fillers to secure continuity of speech flow
(especially wiesz ‗you know‘, cf. Schiffrin‘s (1987) analysis of źnglish y‘know, and rozumiesz
‗you understand‘). In these functions, especially the latter one, the respective elements may be
considered semantically eroded – an observation O óg (1991Ś 185) himself makes as wellŚ
―Nie nale y do rzadko ci, kiedy informacja typu meta przerasta wła ciwą (it is not seldom
that information of the type meta [i.e. the pragmatic information – S.B.] surmounts the proper
4
one)‖4 – which makes a self-correction impossible, as no semantics exist that could be
interchanged for a more corresponding one.
Unfortunately, O óg does not elaborate his self-correction test on other examples of ME with
more specific, maybe even composite semantics, although he treats them in other aspects.
MEs based on mówiąc ‗speaking‘ belong to the latter type and are a good example that such
elements may also be subject to self-correction (2).
(2) Mówiąc krótko, chciałem powiedzieć, (mówiąc) szczerze, Janek nie odpowiada mojej
koncepcji dobrego przyjaciela.
‗In short (lit. shortly speaking), I wanted to say, frankly (speaking), Janek does not
correspond with my notion of a good friend.‘
At the same time, example (2) illustrates quite nicely that O óg‘s self-correction test is in fact
an implicit test for the scope of negation, as (3) is a periphrasis of (2) with overt negation at
work.
(3) Mówiąc nie krótko, a szczerze, Janek nie odpowiada mojej koncepcji dobrego
przyjaciela.
‗Not in short (lit. shortly speaking), but frankly, Janek does not correspond with my
notion of a good friend.‘
Non-negatability is one of the defining properties Rathmayr (1985: 72) proposes for particles
serving as pragmalexemes, i.e. for particles with metatextual function. This criterion has to be
modified regarding the scope of negation. As a particle consists of just one lexical item, it is
not surprising that the (factually inacceptable) negation scopes over the whole particle. MEs,
however, may be constructions (henceforth Cxns) consisting of several lexical items. Quite
interestingly, different MEs behave differently under negation. O óg‘s self-correction tests
the negatability of the whole metatextual element, but even if the MEs consist of several
lexical items, the type of elements discussed by him cannot be subject to partial negation
either: e.g., for the ME mówię ci the negation of the (enclitic) pronominal element *mówię nie
ci, a mu ‗I don‘t tell you, but him‘ is not acceptable, whereas mówię in the predicative
function allows for the negation of its (long, i.e. focused) pronominal argument (mówię nie
furthermore Birzer (2015: chapter 4.5.) on the semantic voidness of ogólnie (mówiąc)
4 cf.
‗generally (speaking)‘ in the filler function.
5
tobie, a jemu ‗I don‘t tell you, but him‘). This is also corroborated by the behavior which is
displayed under negation by powiedzmy ‗let‘s say‘ (Wołk 2007Ś 10), another representative of
the type of Mźs discussed by O óg. Contrary to these findings, the MEs based on mówiąc
‗speaking‘ show that some types of MEs allow partial negation if the modifier is negated (3),
but not if the head (4-6) is under negation.
(4) *Nie mówiąc szczerze, Janek nie odpowiada mojej koncepcji dobrego przyjaciela.
‗Not speaking frankly, Janek does not correspond with my notion of a good friend.‘
(5) *Nie mówiąc, a powiedziawszy szczerze, Janek nie odpowiada mojej koncepcji dobrego
przyjaciela.5
‗Not speaking, but having spoken frankly, Janek does not correspond with my notion of a
good friend.‘
(6) *Nie mówiąc, a pisząc szczerze, Janek nie odpowiada mojej koncepcji dobrego
przyjaciela.
‗Not speaking, but writing frankly, Janek does not correspond with my notion of a good
friend.‘
As we are interested in MEs based on infinite inflectional forms of verba dicendi, we will also
have a look on MEs based on the infinitives powiedzieć ‗say‘ and rzec ‗say‘. Our research in
the Polish National Corpus6 showed that there exist only two constructional types of MEs on
this basis: aby /by / żeby tak powiedzieć / rzec ‗so to say / in order to say so‘ (8) and aby /by /
żeby nie powiedzieć / rzec ‗not to say / in order not to say‘ (7). The fact that aby / by / żeby nie
powiedzieć / rzec ‗not to say / in order not to say‘ already contains a negation, but no
modifier, makes the insertion of a second negation impossible (7i). Structurally, it might
theoretically be possible that aby / by / żeby nie powiedzieć / rzec ‗so to say / in order to say
so‘ allows partial negation, i.e. negation of the modifier, but the semantics of tak ‗so‘
prohibits this (8i). Negation of the head is impossible (8ii), just like with mówiąc.
(7) Szanse są więc niewielkie, by nie powiedzieć zerowe.

‗The chances are thus small, in order not to say zero.‘
5 Some MEs based on the adverbial participle have doublets in the sense that they can be formed either with
the simultaneous adverbial participle ówią or with the anterior one powiedziawszy. Here we test the
possibility of partial negation by using such a doublet.
6 if not indicated otherwise, all corpus research is conducted in the balanced subcorpus containing 300 M
segments.
The query was [orth=",|.|;|:|-" a y| y|że y powiedzie |rze | ówi .
6
(Trybuna ląska, 2004-04-24)
(7i) *Szanse są więc niewielkie, by nie nie powiedzieć zerowe.
‗The chances are thus small, in order not not to say zero.‘
(8) Wyjątkowy ma także życiorys i - żeby tak powiedzieć - osiągnięcia.
‗He [Trotsky‘s murderer] also has an extraordinary CV and – so to say –
accomplishments.‘
(Gazeta Wyborcza, 1993-06-09)
(8i) *Wyjątkowy ma także życiorys i - żeby nie tak, a inaczej powiedzieć - osiągnięcia.
‗He [Trotsky‘s murderer] also has an extraordinary CV and – not to say so, but in
another manner – accomplishments.‘
ii
(8 ) *Wyjątkowy ma także życiorys i - żeby tak nie powiedzieć, a pisać - osiągnięcia.
‗He [Trotsky‘s murderer] also has an extraordinary CV and – so not to say, but write –
accomplishments.‘
To summarize this stage of analysis, MEs behave differently regarding negation. Among
those MEs consisting of several lexical items, some allow partial negation scoping over the
modifier, but never over the head, and others do not allow any negation whatsoever. Formal
criteria do not seem to play a role, as MEs based on infinite mówiąc may take partial negation,
but MEs based on infinite powiedzieć and rzec do not. Instead, the semantics seem to explain
the different behavior of MEs. To get the full picture, the semantic analysis must consider also
another characteristic semantic property of MEs, namely their syntactic eliminability due to
their irrelevance for the content, i.e., the proposition ( 4i and 8iii; cf. Rathmayr 1985: 42).7
(4i) Janek nie odpowiada mojej koncepcji dobrego przyjaciela.

‗Janek does not correspond with my notion of a good friend.‘
(8iii) Wyjątkowy ma także życiorys i osiągnięcia.
‗He [Trotsky‘s murderer] also has an extraordinary CV and accomplishments.‘
Tak ‗so‘ in (8) is a deictic particle referring to (part of) the content, but without any other
semantic components. Just like the elimination of the ME aby /by / żeby tak powiedzieć / rzec
is irrelevant for the content, its partial negation with e.g. inaczej ‗otherwise‘ (8i) also has no
7 This approach also finds a parallel in general linguistics, where Fraser (2006: 189) establishes integration into a
discourse segment without contributing to its proposition as a defining criterion. However, as our discussion of
semantics shows quite nicely, this criterion is a semantic rather than a syntactic one, although it carries syntactic
in its name.
7
influence on the content, but creates a contradiction, since the content offers just one point of
reference for both tak and inaczej – namely osiągnięcia ‗accomplishments‘. The working
mechanism for aby / by / żeby nie powiedzieć / rzec ‗not to say / in order not to say‘ is
basically the same, as it also relies on reference to (part of) the content. The MEs based on
mówiąc behave differently in this respect. Their constitutive lexical elements, such as szczerze
‗frankly‘, have semantics of their own which do not refer to (part of) the content. Therefore,
the negation of these items does not lead to a referential mismatch, but opens a slot for
another lexical item carrying ―semantic weight‖, such as krótko ‗shortly‘. We may
furthermore deduce that partial negatability is a test to establish (semi)productivity in
construction-based MEs such as ADV + mówiąc. The mere possibility of extending the
construction by inserting nie X a Y ‗not X but Y‘ suggests productivity. If the set of elements
eligible for insertion in places X and Y is restricted, we may speak of a partially lexicalized
semiproductive pattern.8
To summarize, our discussion of negatability in general and of O óg‘s self-correction test in
particular shows that negatability is a weak property and needs to be accompanied by other,
more distinct ones. One of them is syntactic eliminability, which interplays with negatability
on the semantic level.
Additionally, Rathmayr (1985) offers some more properties:
MEs cannot be subject to questions, i.e., as with particles (cf. Rathmayr 1985: 72) MEs do not
serve as answers to (probe) questions.9
(4ii) Jakim sposobem / dlaczego Janek nie odpowiada mojej koncepcji dobrego
przyjaciela? – *Szczerze mówiąc.
‗In which way / why does Janek not correspond with my notion of a good friend? –
Żrankly speaking.‘
(8iv) Jakim sposobem / dlaczego ma także wyjątkowy życiorys i osiągnięcia? – *By tak
powiedzieć.
‗In which way / why has he [Trotsky‘s murderer] also an extraordinary CV and
accomplishments? – So to say.‘
8 Fraser s statement that some of the elements he defines as discourse markers form the interchangeable part of
fixed constructions (cf. Fraser 2006: 194) supports our observation on the semiproductivity of some
constructions producing MEs. This, however, does not mean that we share Fraser s concept of discourse
marker.
9 Powiedzmy ‗let‘s say‘ seems to be an exception from the rule, as it may serve as answer at least to polar
questions (Wołk 2007Ś 7).
8
Rathmayr (1985: 72) gives two more defining criteria typical for particles, namely that they
are unstressed and cumulative with other particles. However, these apply only partially to
MEs. MEs may bear secondary stress if they consist of several words.
Just as well, the possibility of using MEs cumulatively also varies. One dividing line seems to
run between spoken and written language: (1) is a transcript of spoken language, where MEs
with the same function – direction of hearer‘s attention – are used cumulatively. In written
language, probably due to stylistic reasons, cumulation seems to be possible in two cases. In
the first case, the MEs serve different discoursive needs, as shown in (9) below. Zresztą ‗by
the way‘ is a text structuring device, whereas ci lej mówiąc ‗more precisely speaking‘ may
be considered a reformulation marker.
(9) Takim [zwierzęciem krwiożerczym – S.B.] mnie stworzono. Zresztą ściślej mówiąc, nie
tyle krwiożerczym,ile mięsożernym ...
‗I was created as such a bloodthirsty beast. By the way, more precisely speaking, not
so much as a bloodthirsty, but a carnivorous beast …‘
(Sł. Mro ek. 1997. Teatr 3.)
Concretization constitutes the second case and seems to involve only MEs with connective
function, usually reformulation markers. Mendoza (2009: 983) states that
[d]ie Bedeutung bzw. Funktion von multifunktionalen koordinierenden Konjunktionen

kann durch die Kombination mit sog. Konkretisatoren spezifiziert werden […], wobei
die Konkretisatoren alleine wiederum meistens ebenfalls Konnektorenfunktion
übernehmen können.
[the meaning, or the function of multifunctional coordinating conjunctions,
respectively, may be specified by a combination with so-called concretisators […],
whereupon in most cases the concretisators may, in turn take over the function of
connectives. translation – S.B.]
Czy in example (10) is a multifunctional item that may serve as a complementizer with the
meaning ‗whether‘, as a conjunction with the meaning ‗or‘, and as a question particle.
Wła ciwiej mówiąc ‗more correctly speaking‘ specifies that czy functions as a conjunction.
9
(10) Ten wybór, czy właściwiej mówiąc nominacja [sic!], [...] Gierek [...] przyjął bez
skrępowania.
‗This election, or more precisely speaking the nomination żierek accepted without
hesitation...‘
(J. Rolicki. 2002. Edward GierekŚ życie i narodziny legendy.)
Now that the general criteria for identifying MEs have been discussed, some words are in
order on our choice of MEs based on infinite forms of verba dicendi as research object. This
choice shares all the semantic traits that have been described above as typical for MEs based
on verba dicendi and will henceforth be referred to as KMs. Let us reconsider abowska‘s
(2009: 181) properties of metatextual commentsŚ ―występują na powierzchni tekstu w postaci
skróconych zdań, najczę ciej zwrotów imiesłowowych, […]ś mogą to być wyra enia
samodzielne, wkomponowane w składnię wypowiedzenia, wyra enia wtrącone, występujące
w konstrukcjach tzw. wyodrębnionych [occur on the text surface in the form of reduced
clauses, most frequently in the form of adverbial participle clauses […]ś they can be
autonomous expressions integrated into the syntax of the utterance or parenthetical
expressions figuring in so-called detached constructions – translation S.B.]‖. At first sight, it
seems that all the MEs based on verba dicendi that are discussed by O óg (1991) fulfill these
criteria, since mówię ci ‗I tell you‘, wiesz ‗you know‘ or rozumiesz ‗you understand‘ may be
considered reduced clauses, in which the syntactic realization of the semantic role ‗content‘ is
missing. However, another possible analysis reads the utterance, i.e. the actual information, as
realization of the semantic role ‗content‘. This is the more probable, as the pragmatic function
of O óg‘s MEs enumerated above is guiding the hearer‘s attention to the following
information. Admittedly, the syntactic realization does not correspond to the prototypical
scheme expected in written language, where the utterance content forms a complement
sentence, but structures of the type mówię ci (to)Ś CONTENT X ‗I say to you (this)Ś content X‘
or co ci powiemŚ CONTENT X ‗I will tell you somethingŚ content X‘ are also acceptable in
written language and parallel the encoding of direct speech in both written and spoken
language. The specifics of spoken language, where intonation contours and pauses structure
information make such constructions even more expectable and acceptable. This means that
the enumerated expressions do not fulfill abowska‘s criterion of integration into the
utterance, especially if they are set off from the actual information by a pause or two different
intonation contours stretching over the ME and the utterance proper respectively. The former
has been shown to be the case e.g. for the parenthetical ME powiedzmy ‗let‘s say‘ (Wołk
10
2007: 7). In contrast, expressions based on adverbial participles of verba dicendi constitute
prototypical representatives of KMs, as abowska (2009Ś 181) states herself in the citation
above.
Another evidence for absent integration into the clause is the positioning of the enclitic
conditional marker by + ending. This complex is known to be attached to the verbal stem only
loosely and to cliticize preferably to the ―first pronounced word in the clause‖ (Hansen 2010:
345). Consequently, if MEs were integrated into their host clause, the conditional marker
should cliticize to them if they take sentence-initial position. Our corpus research for the ME
wiesz ‗you know‘, and the KMs based on the infinitive powiedzieć ‗say‘ and rzec ‗say‘ as
well as the adverbial participle mówiąc ‗speaking‘ showed that this is not the case.10
Furthermore, just like KMs based on the infinitive of verba dicendi, the KMs based on the
adverbial participle display morphosyntactic properties that distinguish them clearly from
other occurrences of verba dicendi, as the discussion below will show. Additionally,
constructions containing an adverbial participle are a (semi)productive means to form KMs
(cf. Birzer accepted a), whereas other verb forms seem to be used only in haphazardly
lexicalized units. Therefore we limit our discussion of KMs based on verba dicendi to those
with the adverbial participle or infinitive as formant.
Morphosyntactically, our KMs are distinct from other occurrences of verba dicendi in the
following respects: For the adverbial participles, undoubtedly the most important syntactic
issue is the loss of co-reference between the covert subject of the adverbial participle and the
first argument of the matrix verb. Instead, the speaker is the covert subject of the adverbial
10 we assumed that the clitic does not take intermediate position between the elements of the KMs based on
verba dicendi. Therefore, we conducted the queries
"wiesz|rzec|powiedzieć|mówiąc" by for cases without punctuation
"wiesz|rzec|powiedzieć|mówiąc" ",|Ś" by for the cases with punctuation
We found four instances, where the conditional marker by + ending follows immediately after the KM
constituent mówiąc ‗speaking‘. However, in (1) the KM does not constitute the first word in the sentence, so we
may interpret (1) as evidence that the KM forms one prosodic segment with the element it scopes over, i.e.
najchętniej ‗preferably‘, and the conditional marker by cliticizes to this prosodic segment. (2) represents the
other three instances, where the KM is followed by purpose clause introduced by the (contracted)
complementizer by.
(1) Najchętniej, szczerze mówiąc, bym słyszał pańskie odpowiedziŚ
‗Preferably, frankly speaking, I would like to hear your answersŚ‘
(Stenogram z 53. posiedzenia Komisji ledczej do zbadania ujawnionych w mediach zarzutów dotyczących przypadków
korupcji podczas prac nad nowelizacją ustawy o radiofonii i telewizji (SRTV) w dniu 2 lipca 2003 r.)
(2) Ikona próbuje czynić widzialnym niewidzialne, pozwala tym samym, aby widzialne nie przestawało odsyłać do czego
innego niż ono samo. Inaczej mówiącŚ by niewidzialne nie zastygło w widzianym.
‗The icon tries to make the invisible visible, and allows in the same way that the visible does not stop to relate to
something else than just to itself alone. In other words (lit. in another way speaking): that the invisible does not congeal
in the visible.‘
(Dariusz Czaja. 2009. Lekcje ciemno ci.)
11
participle (11). The aforementioned co-reference is the characteristic trait of prototypical
adverbial participles (12).
(11) Øsp Szczerze mówiąc, Adam jest głupi.

frankly speak-AP Adam-NOM is stupid-NOM.SG.M
‗Żrankly speaking, Adam is stupid.‘
(12) Øj Mówiąc z kolegą, Adamj się miał.
speak-AP with friend-INSTR Adam-NOM REFL laugh-PST.3SG.M
‗Speaking with a friend, Adam laughed.‘
The loss of co-reference also indicates that the adverbial participle ceases to function as
secondary predication. It is a trait that all adverbial participles share – even those that may not
be considered prototypical due to the co-reference of their covert subject with a matrix verb
argument other than the first one (13).11
(13) Wujek prosi nas zwrócić pożyczone pieniądze, nie mówiąc o tym nikomu.
‗The uncle aks us to return the borrowed money, not speaking with anybody about it.‘
(cited after Feret 2005: 83)
For the KMs based on the infinitive, loss of co-reference and control is the most important
issue. Hints to be discussed below imply that the point of departure for this kind of KM is a
purpose clause or complement clause introduced by the complementizer (że)by ‗in order toś
that‘ and containing an infinitve as predicative element (14). The infinitive is controlled by
the first argument of the superordinate clause (14; cf. Hansen 2010 for reference and control
in Polish conditional and purpose clauses). If the construction is used as KM, however,
control by and referential identity with the first argument of the superordinate clause are
replaced by control by and referential identity with the speaker, who becomes covert first
argument of the infinitive (15).
(14) Zadzwoniłamj, żeby ci powiedziećj,

call-PST.1SG.F COMPL you-DAT say-INF
że znalazłam przyjaciółkę dla twojego psa.
11 in fact, the loss of secondary predication status implies that the affected item should no longer be labelled as
adverbial participle. Yet for reasons of convenience we will retain this term.
12
that I found a girlfriend for your dog
‗I called you in order to tell you that I found a girlfriend for your dog.‘
(Filipiak, I. 2006. Magiczne oko. Opowiadania zebrane.)
(15) Sytuacja jest niewesoła, żeby nie powiedzieć –
situation-NOM is not_funny-NOM.SG.F COMPL not say-INF
dramatyczna.
dramatic-NOM.SG.F
‗The situation is not funny, not to say – dramatic.‘
(Tygodnik Ciechanowski, 2002-09-16)
Why do we assume the purpose clause as point of departure for the items under investigation?
The secondary complementizers aby ‗in order to‘ and żeby ‗thatś in order to‘ consist of the
primary conjunction a ‗and, but‘ or complementizer że ‗that‘respectively and the conditional
marker by. They introduce purpose clauses as well as one complement type of control verbs.
In both cases, the controller reference is decisive:
―In complement clauses governed by [control] verbs, the conditional is used if the subject of the main clause does not
control the subject of the subordinated clause, i.e. if the subjects refer to different entities.[…] In the case of referential
identity the conditional is excluded and we have to use the infinitive [as direct argument of the control verb, instead of
a complement clause – S.B.]‖ (Hansen 2010: 352)
The same rule applies to purpose clauses (Hansen 2010: 353). Since our KMs have the
speaker as covert first argument, this means that as the point of departure for our KMs we
must assume purpose clauses or complement clauses of control verbs whose first argument
was co-referent with the first argument – in the first person (singular) – of the superordinate
clause.
Some words are in order on why we do not include participles into our analysis, although they
undoubtedly constitute infinite verb forms. Agreeing participle forms can be generally
excluded, as they are integrated into their host utterance and thus offend against one of our
defining criteria for KMs. Therefore, only the indefinite-personal active predicates on –no/-to
(henceforth –no/-to forms), which have developed out of the passive participle via diathesis
from passive to active (cf. Weiss 1984: 156), remain for discussion. Functionally, they rather
remind finite forms, but they are morphologically infinite and their representatives formed of
13
verba dicendi, namely mówiono ‗having spoken‘ and powiedziano ‗having said‘12, have the
ability to function as predicates of parenthetical Cxns in collocations of the type jak mówiono
/ powiedziano (powyżej) ‗as has been said (above). However, our corpus analysis told a
slightly different story.13 In a first step, we searched14 for adverbs that colligate with the
described –no/-to forms in order to identify adverbial collexemes. 15 The results showed that
adverbs – even jak ‗so, as‘ and (po)wyżej ‗above‘ – function as collexemes only rather
seldomly16 and that the construction as a whole lacks several of the characteristic traits of
KMs.
12 as English has no corresponding construction, we will make use of the passive construction in the
translations of our examples. Rzeczono ha i g said is ot dis ussed, as it as attested o l t i e i the
corpus. The pa ti iple ze zo afo e e tio ed is not analyzed because it integrates into the utterance and
thus does not fulfill our criteria of KM.
13 if not indicated otherwise, research has been conducted in the balanced subcorpus of NKJP.
14 for these ends we used the queries
[pos=adv] mówiono [] [orth!= e]
[pos=adv] powiedziano [] [orth!= e]
mówiono [pos=adv] [] [orth!= e]
powiedziano [pos=adv] [] [orth!= e]
that excluded instances with complement clauses introduced by że ‗that‘, as they are indicative of the
predicative usage of verba dicendi.
15 for the notion collexeme se Stefaenowitsch & Gries 2003.
16 this finding is corroborated by an additional search for the collocation of all inflectional forms based on the
participle passive stems powiedzian- and mówion- and the adverb (po)wyżej a o e (queries powyżej
powiedzian.* respectively mówion.* and vice versa) gave just 17 matches: no matches were attested for
mówion.*, and of the 17 matches for powiedzian.* only seven represented the impersonal construction on –o
(the others are instances of participles, which integrate into the utterance and thus do not agree with our
criteria of KM). Unfortunately, the contexts were not large enough to reconstruct whether the speaker or a
protagonist from the narration has to be assumed as demoted agent. These findings are the more interesting,
as powyżej a o e o u s ti es i dis ou se st u tu i g o te ts (query [base=powyżej & pos=adv],
followed by manual post-processing); however, these contexts do not display signs of (ongoing) lexicalization:
(1) Sentymentalizm - w całej jego powyżej przypominanej rozciągło ci - jest bliźniakiem kiczu i jeden bez
drugiego żyć nie może.
‗Sentimentalism – in its whole extension mentioned above – is the twin of kitsch and one cannot live
without the other.‘
(Stanisław Mro ek. 1975. Jak zostałem filmowcem)
(2) Znowu [to zrobiłem – S.B.] nie z sympatii, ale z motywów przedstawionych powyżej.
‗Again I did this not out of sympathy, but out of the motives presented above.‘
(Stanisław Mro ek. 1975. Jak zostałem filmowcem)
(3) ... redakcja chętnie oceni Twój tekst – jednak, z powodów jak powyżej, przypomnij jej się tak raz na miesiąc
‗The editorial board will be happy to assess your text – however, for the reasons as above, remind us about it
roughly once in a month.‘
(Archiwa magazynu Esensja. 2005-2008).
(4) Powyżej pisałem o cudownej kreacji jego córki, ale kilka razy naprawdę niewiele zabrakło do tego, by
ukradł jej cały show.
14
The functional core of each KM lies in its metatextual orientation, and the construction under
investigation does not meet them (with some rare exceptions). Although our queries excluded
instances with complement clauses, the majority of all matches still took complements, e.g.
(16-17).
(16) [Budynek teatru niemieckiego – S.B.] był symbolem szowinizmu, a z jego sceny, jak
powiedziano przy otwarciu, nie miało nigdy pa ć słowo polskie.
‗The building of the żerman theater was a symbol of chauvinism, and on its stage, as was
said at the opening ceremony, a Polish word should never ever be uttered.‘
(Jan Bałdowski. 1997. Warmia, Mazury, Suwalszczyzna.)
(17) Niektórzy już tacy są, że potrzebują wodza. – A Putin im nie wystarczy? – Nie
wystarczy. Putin przy Stalinie to liliput. Albo - jak mówiono o Breżniewie - niedonosek.
‗Some are already of the kind that they demand a vozhd‘. – And Putin does not suffice
them? – No, he doesn‘t. In comparison to Stalin, Putin is a liliput. Or – as was spoken
about Brezhnev – a preemie.‘
(Aleksander Kaczorowski. Włodzimierz Wojnowicz, pisarz rosyjski, o Polsce, źuropie i
nowej powie ci. // Polytika. 2006-04-01)
As described above, the existence of complements constitutes one of the dividing lines
between ―regular‖ inflectional forms and the verbal constituents of KMs. Another
distinguishing feature is the speaker as (demoted) first argument of the KM. However, most
instances of the –no/-to construction without complements do not allow for this interpretation,
be they modified by an adverb (18) or not (19-21). Rather, folks or the man on the street have
to assumed as demoted first argument; in (19) – one of the previous speakers in the
discussion.
(18) Zwłaszcza w ciągu ostatnich trzech lat przed wojną władze państwowe po więcały
sprawie budowy wygódek (albo sławojek, jak mówiono ironicznie) szczególną uwagę.
‗źspecially in the course of the last three years befor the war, the government paid special
attention to the construction of sanitary facilities (or jakes, as was said ironically).‘
Above I have written about his daughter‘s wonderful creation, but there were several occasions when not
much was missing and he would have stolen her the whole show. ‗
(Michał Hernes. Matematyka mo e być fascynująca! // Esensja, 6/2008)
15
(Stanisław Berenda-Czajkowski, 2001. Dni grozy i łez.)
(19) Pan marszałek Płażyński był uprzejmy nie poddać go [wniosek – S.B.] pod
głosowanie, i słusznie, ponieważ wniosek był nie tyle nieprawidłowy, jak powiedziano, czy
niezgodny z regulaminem, ile przedwczesny.
‗Chairman Pła yński was so kind as not to make the petition subject to voting, and rightly
so, because the petition was not so much contrary to rules, as has been said, or
incompatible with the regulations, than the previous one.‘
(Sprawozdanie stenograficzne z obrad Sejmu RP z dnia 11.04.2001, 3 kadencja, 106
posiedzenie, 2 dzień.)
(20) Welch sprzedał bank - jak mówiono – "po cenie złomu".
‗Welch sold the bank – as is said – ―for the price of scrap‖.
(Gazeta Wyborcza. 1997-06-06)
(21) Zaczyna się budowanie nowej Rosji, a wszystko dzięki Putinowi - przekonują
kremlowscy politycy i większo ć Rosjan przyjmuje te zapewnienia z pełną wiarą.
Zwłaszcza że słyszą to od witu do nocy, jak w dawnym dzienniku telewizyjnym z czasów
Breżniewa, w którym - jak mówiono – były wyłącznie wiadomo ci o Leonidzie Iljiczu i
trochę informacji o pogodzie.
‗The construction of a new Russia is starting, and all this thanks to Putin – assure the
Kremlin politics and the majority of Russians accept these promises with full faith. The
more so, as they hear this from dusk till dawn, like in the former TV news broadcasts in
Brezhnev‘s times, where there were – as was said – exclusively news about Leonid Ilich
and a little bit of information about the weather.‘
(Sławomir Popowski. Obywatele, głowa do góry. // Polityka, 2007-12-01)
Only in highly formal texts, such as the statement from the Supreme Court, the modified
construction based on mówiono or powiedziano respectively functions clearly as KM (22).
(22) ... zarzut podnoszony przez oskarżonego nie dotyczy sądu wła ciwego, lecz wyraża
obawę o kierowanie się solidaryzmem rodowiskowym. W takiej jednak sytuacji, jak
powiedziano, każdy inny sąd może narazić się na analogiczny zarzut.
‗... the objection raised by the accused does not relate to the specific court, but expresses
the concern about being guided by solidarity with one‘s own milieu. However, in such a
situation, as said, any other court might be confronted with a similar objection.‘
16
(POSTANOWIENIE Z DNIA 17 SIERPNIA 2004 R., V KO 47/04. Orzecznictwo Sądu
Najwy szego, Izba Karna. 2004-08-17)
For these reasons we decided to refrain from discussing the construction comprehensively.
2. Functions in Contemporary Polish

Since the functions of KMs based on the adverbial participle mówiąc have already been
described in Birzer (accepted a), we limit ourselves to a short survey of the three
superordinate functions and the construction that prototypically goes along with the respective
function.
The first superordinate function is the stance-marking function with ADV + mówiąc as
prototypical construction. This construction has 104 representatives, i.e. in the Polish National
Corpus 104 different adverbs are attested to fill the ADV slot of this construction, among them
the near-synonyms delikatnie / łagodnie mówiąc ‗mildly speaking‘, dokładnie / konkretnie /
precyzyjnie mówiąc ‗precisely speaking‘ and generalnie / ogólnie mówiąc ‗generally
speaking‘. On the pragmatic level, the construction (henceforth Cxn) ADV + mówiąc expresses
speaker‘s stance towards the proposition. On the syntactic level, the Cxn may function as
connective. Both functions are inherent in the Cxn, but depending on the semantics of the
adverb inserted into the Cxn, their manifestation varies. Thus, connectivity is less pronounced
with e.g. szczerze mówiąc ‗frankly speaking‘, but speaker‘s stance figures prominently
(compare (23), where speaker‘s stance is dominant, with (24), where connectivity is also at
work); whereas with e.g. krótko mówiąc ‗in short‘ connectivity is dominant (25) and speaker‘s
stance plays only a minor role (compare (25) with (26), where stance is more in focus). A case
study of the various syntactic contexts and thus the various semantic and functional nuances is
offered for ogólnie mówiąc ‗generally speaking‘ in Birzer (2015).
(23) Malownicza zakopiańska zima przerodziła się w wiosenne roztopy. Szczerze mówiąc,
wolę skrzypiący nieg pod stopami niż błoto ...
‗The picturesque winter in Zakopane degraded into springtime thaws. Żrankly speaking, I
prefer crunchy snow under my feet over mud…‘
(Mój "Dzienniczek Polski" // Dziennik Polski, 2002-02-01)
(24) Wystąpienie było bardziej nastawione na efekt propagandowy niż merytoryczny - z
podaniem konkretnych spraw czy terminu rozwiązania zagadnień. Szczerze mówiąc,
17
informacja była przypomnieniem albo powtórzeniem exposé wygłoszonego przez pana
premiera ...
‗The address was aimed rather at a propagandistic effect than at content – with the
presentation of concrete issues or a date for the solution of issues. Frankly speaking, the
information was a reminder or repetition of the exposé that had been delivered by the
prime minister … ‗
posiedzenie, 3 dzień)
(25) ... ponad 64% przestępstw popełnianych w naszym kraju nie jest znanych policji [...].
Wystarczy powiedzieć, że według cytowanych badań policja nie wie o ok. 70% pobić, 80%
kradzieży i ř0% przestępstw o charakterze seksualnym. Krótko mówiąc, w wietle badań
wiktymizacyjnych Polska jest krajem, w którym zagrożenie niektórymi typami przestępstw,
zwłaszcza popełnianych przy użyciu przemocy, jest jedno z najwyższych w wiecie.
‗... more than 64% of the crimes committed in our country are not reported to the police
[…] It‘s sufficient to state that according to the cited police surveys the police don‘t know
of about 70% of all batteries, 80% of all thefts and 90% of all crimes with sexual
character. In short, in the light of victim surveys, Poland is the country where the threat of
several types of crimes, especially of those committed with violence, is among the highest
in the world.‘
(26) Halina ma płaszcz. Taki płaszcz to skarb, choć na pozór nic specjalnego - znoszony,
trochę za długi i zbyt szeroki, krótko mówiąc, niemodny.
‗Halina has got a coat. Such a coat is a treasure, although to all appearances it is nothing
special – worn out, a bit to long and a bit to wide, in short, unmodern.‘
(Granica wytrzymało ci. // Dziennik Polski. 2006-06-03)
Quite interestingly, the Cxn ADV + mówiąc forms a doublet with ADV + powiedziawszy, i.e.
with a Cxn based on the perfective adverbial participle with anterior meaning. Yet the latter
Cxns has to be considered a non-productive pattern for forming KMs, since only seven
different instantiations – with the adverbs dokładniej ‗more precisely‘, inaczej ‗differently‘,
lepiej ‗better‘, najpro ciej/prosto ‗(most) simply‘, ci le(j) ‗strictly‘ and szczerze ‗frankly‘
and the noun prawdę ‗truly (lit. the truth)‘ – are attested in the Polish National Corpus.
18
The second superordinate function is contextualization with the prototypical Cxn AP +
ADJ.INSTR + NOUN.INSTR, for which Polish offers 27 instantiations. It relates (part of the)
proposition within a certain discourse – be it modern terminology for well-known phenomena
(cf. (27) or different style registers that characterize certain types of discourse (28).
(27) Dziwne zjawiskoŚ skrajna prawica i skrajna lewica zgadzają się w sprawach granic -
no, może nie całkiem, ale tak na ř5%. Zgodnie też proponują plan, mówiąc dzisiejszym
językiem, czystek etnicznych ...
‗A bizarre phenomenonŚ The ultimate right and the ultimate left agree in [national – S.B.]
border issues – well, maybe not completely, but to perhaps 95%. In unison they also
propose a plan of, speaking in today‘s language, ethnic cleansing …‘
(Andrzej Anonimus. 1999. Nie nadaje się, przecież to jeszcze szczeniak.)
(28) Skuteczny zrazu sprzeciw pozwanych [...] bazował na "braku czynnej legitymacji
prawnej po stronie powodów". Mówiąc to samo językiem kolokwialnym: "Nie do
przyjęcia jest, że czego chcą jacy Jabłkowscy, będący spadkobiercami niektórych tylko
wła cicieli, nieistniejącej od daywna firmy".
‗The effectful objection of the defendants [...] based on the ―absence of an effective legal
legitimation on the complainants‘ side‖. Saying the same in colloquial languageŚ ―It
cannot be accepted that these Jabłkowskis want something, who are the heirs to some
proprietors of a for long non-existent company.‖‘
(Żeliks Jabłkowski. 2005. Dom Towarowy Bracia JabłkowscyŚ romans ekonomiczny.)
The third superordinate function is the marking of quotation17 (29) with the prototypical Cxn
mówiąc + językiem/słowami.INSTR + NOUN.GEN. The Cxn provides ―discourse-salient
information‖ (żüldemann 2012Ś 140), since it lays open the source of the discourse rendered
(cf. also Weiss in preparation, who states for the usage of citations in political discourse that
―the identification [of the source cited – S.B.] is the more explicit in cases where the source
cannot be considered to be common knowledge‖ [translation – S.B.]). At the same time, it
also links two discourses with each other, namely the discourse into which the citation is
integrated to the one from which the citation was taken. Weiss (in preparation) describes the
mechanism at work as follows: As basis for the connection [between citation and the text it is
17 Quite interestingly, Wołk notes a ―citational‖ usage (użycie cytacyjne) of powiedzmy (2007: 13-15).
However, this citational usage does not encompass a function from the quotative realm, but rather the fact that
the truth value of the utterance segment marked by powiedzmy cannot be determined.
19
integrated into – S.B.] functions the metatextual operation of comparison, i.e. an implicit or
explicit parallel between the actual […] situation […] and the content of the xeno-text [the
quote – S.B.]‖ [translation – S.B.].
(29) Ja my lę, że obaj mają swoje - mówiąc językiem Lecha Wałęsy - plusy dodatnie i plusy
ujemne.
‗I think that both have their – using Lech Wałęsa‘s wording (lit. speaking with Lech
Wałęsa‘s language) – positive pluses and negative pluses.‘
(Dziennik Polski. 2001-05-18)
Let us now turn to the functions of the Cxns based on the infinitives of verba dicendi. These
Cxns consist of three elements – a complementizing element, the negation nie ‗not‘ or the
deictic element tak ‗so‘ and an infinitive (30). The choice of the ―intermediate‖ element
determines the function of the Cxn; the complementizers and infinitives are basically
interchangeable.18 Quite interestingly, our corpus research showed that only the perfective
verbs powiedzieć ‗say‘ and rzec ‗say‘ serve as bases for these KMs.
(30)
żeby tak ‗so‘ powiedzieć ‗say‘
by nie ‗not‘ rzec ‗say‘
aby
Aby / by / żeby tak powiedzieć / mówić / rzec ‗so to say / in order to say so‘ has two functions.
The first one is that of a downtoning hedge:
A hedge is either defined as one or more lexico-syntactical elements that are used to modify a
proposition, or else, as a strategy that modifies a proposition. The term 'hedging' is used to refer to the
textual strategies of using linguistic means as hedges in a certain context for specific communicative
purposes, such as politeness, vagueness, mitigation, etc.
(Schröder and Zimmer 2000 cited after Gries / David 2007)
The semantic implications of the constructional element tak ‗so‘ are pivotal for the semantic
explication of aby / by / żeby tak powiedzieć / mówić / rzec in the hedging function, which
18 in our eyes, the fact that Cxns with aby that a e i ge e al u h less f e ue t tha ith że y or by explains
why a y ie ówi ot to sa a d a y tak powiedzie so to sa a e ot attested, a d ot the i o pati ilit
of elements.
20
could sound as follows: ‗Apart from description P, there exist several more ways of
describing the given situation. The speaker chooses P, although they are aware of the fact that
P might be offensive to the hearer and that another description might be more acceptable to
the hearer.‘ Consequently, the hedge scopes over entities – usually autosemantics or NPs
containing autosemantics – that either bear a subjective semantic component in the sense of
Traugott & Dasher (2002: 96) or whose semantics form part of a scale in the wider sense.
E.g., biograficznie ‗biographically‘ in example (31) is part of the scale ―biographical –
fictional‖ and is accompanied by the downtoning hedge by tak powiedzieć, as other
approaches to the interpretation of the drama are also possible. The author of (32) is aware
that the mentioned societal justification of the mentioned reason can be challenged (the
possible evaluation scale ranges from socially justified to unjust); in (33) the downtowner
preceding czynno ciowy ‗functional‘ (33) marks that the adjective is an approximation to the
concept which the author cannot describe more appropriately.
(31) Chmielowski interpretuje go [aktu II dramatu Wyspiańskiego – S.B.] — by tak

powiedzieć — biograficznie, tzn. uważa, iż jest on bezpo rednim przekazem wewnętrznych
walk Wyspiańskiego, jego "spowiedzią autorską".
‗Chmielowski interpretes it [the second act of a drama by Wyspiański – S.B.] – so to say –
biographically, i.e. he pays attention to question whether it is an immediate renarration of
Wyspiański‘s inner struggle, his ―confession as an author‖.‘
(żłowiński, M. 1997. Ekspresja i empatia.)
(32) Jest to nawet racja, żeby tak rzec, "społecznie słuszna".
‗This is even a reason that is, so to say, „justifiable for society‖.‘
(33) To samo wyrażone językiem, aby tak rzec, czynno ciowym, będzie brzmiało
następująco:
‗That same [content – S.B.] expressed in – so to say – functional language, will sound
as followsŚ‘
(Gazeta Ubezpieczeniowa, 2001-09-19)
The second function of Cxns with the element tak ‗so‘ is that of a filler (34-27), i.e. an
element used to cover speech disfluency. It may be considered semantically void, as it does
not contribute any metatextual information; its function is the mere maintenance of speech
flow.
21
(34) Za mały byłem, by tak powiedzieć. Teraz jestem za duży.
‗I was too small, so to say. Now I am too big.‘
(Kornaga, D. Gangrena. 2005)
(35) Wyszedłem, żeby tak powiedzieć, w niedzielę na miasto...
‗I went, so to say, on Sunday to town …‘
(My liwski, W. 2007. Traktat o łuskaniu fasoli.)
(36) Grupa składa się z trzech, żeby tak powiedzieć, podgrupŚ olsztyńskiej, łódzkiej i
żyrardowskiej ....
‗The group consists of three, so to say, subgroupsŚ the ones from Olsztyn, Łód and
irardów … ‗
(Polityka, 2001-12-01)
(37) Niby nie czujesz fizycznego bólu, ale wiesz, że umierasz, i to wywołuje niedający się z
niczym porównać strach. To ten strach boli cię, by tak rzec, naprawdę.
‗You kind of don‘t feel physical pain, but you know, that you are dying, and that
evokes a fear that cannot be compared to anything. And it is that fear that hurts you, so
to say, in real.‘
(Mariusz Czubaj. 2010. 21:37)
As Gries & David (2007) note, many elements used as hedges may also be employed as
fillers. This is probably due to the fact that attenuating expressions usually do not arouse
unpleasant reactions on the interlocutor‘s side, leading to excessive use.
The discrimination of hedging and filler function poses some problems, but is not as unclear
as Gries & David assumeŚ ―Without hearing the utterance, it is not fully clear whether there is
in fact a break around sort of that indicates that sort of is just a disfluency marker.‖ (2007:
note 4) In fact, two characteristics distinguish the filler from the hedging function. The above-
mentioned semantic voidness constitutes the first one and constitutes the basis for the second
one: a filler may occur in any syntactic position (cf. especially 35 and 37), because it is
semantically void and does not dictate any semantic and syntactic prerequisites to its
surrounding. This distinguishes it from all other functions of KMs that are tied to a certain
syntactic (and partially semantic) context.
The negation nie ‗not‘ is the constitutive element for the other group of KMs based on the
infinitives of verba dicendi. The functions of aby / by / żeby nie powiedzieć / rzec ‗not to say /
in order not to say‘ can be ascribed to the superordinate function of stance marking. The KM
22
introduces a reformulation, but at the same time has a downtoning function.19 The KM
implies that the speaker is aware of the fact that the chosen reformulation is probably
conceived as too strong, since the reformulation constitutes a pointed description of the
situation outlined more moderately before.
(38) Refleksja ta otwiera paradoksalne (by nie powiedzieć absurdalne) pole niczym nie
skrępowanej wolno ci interpretacyjnej.
‗This reflection opens a paradox (not to say absurd) field of unlimited interpretative
liberty.‘
(Jerzy Adamski. 2000. wiat jako niespełnienie albo Samobójstwo Don Juana.)
(39) ... i ten prosty, by nie powiedzieć: prymitywny, rachunek pokazuje prawdę ...
‗... and this simple, not to sayŚ primitive calculation shows the truth ...‘
(40) To jednak tylko teoria, aby nie rzec: utopia.
‗But this is only theory, not to say: utopia.‘
(41) Był postawny, by nie rzec gruby, włosy miał lekko przerzedzone, a na twarzy
dobrotliwy u miech.
‗He was stately, not to say coarse, his hair was slightly thinning, and on his face he had a
good-natured smile.‘
(Iwona Surmik. 2005. Ostatni smok.)
What could be the reason that only the perfective infinitives occur in these KMs? The most
plausible explanation seems to be that hedging is central to these KMs, and hedging always
pertains to a specific situation, i.e. hedging is situation-specific. Specific situations are usually
episodic ones, and episodicity is marked by perfective aspect (cf. Hansen 1996).
3. Historical development
The analysis of synchronic data has shown that the syntactic integration of the elements under
inspection must have been dissolved, i.e. the adverbial participles mówiąc and powiedziawszy
have lost the status of secondary predicate, and the infinitives powiedzieć and rzec are no
more part of a subordinate clause controlled by an argument of the superordinated clause. The
19 No instances where just one function is realized has been attested in the Polish National Corpus.
23
speaker has become first argument of all discussed items. Consequently, the first research
question concerns the morphosyntactic and semantic processes that led to the changes
described above. Secondly, KM in Contemporary Polish base on infinite inflectional forms of
just thre verba dicendi – mówić, powiedzić and rzec. Therefore the question arises whether
there existed more competing constructions based on other verba dicendi in the history of
Polish. Furthermore, the historical development of similar constructions in Russian (cf. Birzer
2012 a; b) and Croatian (cf. Birzer accepted c) shows that language contact exerted decisive
influence on the development of these structures, as it resulted in the emergence of a
borrowed construction (Russian borrowed the construction ADV + INF from French and
Croatian the construction ADV + PTCP.PASSIVE from German). Thus it should be verified
whether language contact played a similar role in the history of Polish KMs.
3.1. Used data

Like the synchronic part of this paper, the diachronic one also bases on corpus data.
Therefore, some words are in order on the delimitation of the research object and the corpus
data used.
As one of our research questions concerns competing verba dicendi and inflectional forms
thereof in the historical development of KM, another important issue is identifiying the verba
dicendi to be searched for in the collected data. Since the contemporary KMs are based on
inflectional forms of mówić ‗say / tell, ‗powiedzieć ‗say‘ and rzec ‗say‘, it goes without saying
that these verbs were searched for in the historical texts. For identifying diachronically
competing forms of other verba dicendi, neither introspective data from present-day speakers
nor information from synchronic dictionaries of synonyms are of help, as they do not
incorporate verba dicendi that have gotten out of use, but played some role in the history of
Polish. Therefore we decided to draw on historical bilingual dictionaries, more precisely on
the Latin-Polish dictionaries of Bartłomiej z Bydgoszczy (Kędelska et al. 1999-2009),
Franciscus Mymerus (1997) and the Lexicon Latino-Polonicum (Mączyński 1973). There we
searched for Polish translations of the following Latin verba dicendi:
– dicere ‗say‘ as the probably most versatile Latin verbum dicendi that forms all kind of
inflectional forms and may occur in a wide range of contexts;
– loqui ‗say‘, a deponens that is consequently only able to form semantically active
forms and thus restricted in its contextual usage;
– inquit say-PRS.3SG, a lexicalized form used predominantly for the quotation of direct
speech;
24
– ait say-PERF.3SG, the preterite counterpart of inquit, to be used in the same contexts.
From these sources we elicited powiedzieć ‗say‘, mówić ‗say, speak‘, rzec ‗say‘ and rzekać
‗say, speak‘ as the verba dicendi to be searched for.
Since KMs are rather infrequent items in comparison to other linguistic (core) structures, a
huge data massif is necessary to gather a sufficient amount of instances. Therefore, we
combined research in several media.
The Słownik polszczyzni XVI wieku comprises all lexemes attested in the 16th century, but
suffers from two major drawbacks: firstly, the edition of the dictionary is still going on and
the last available volume covers the lexemes up to Ro-, i.e. rzekać and rzec are not yet
encompassed. Secondly, the Słownik offers several instances for each lexeme and wordform,
but nonetheless the amount of data is still insufficient to reconstruct the development of KMs,
the more so, as not instances of prototypical usage, but bridging contexts are of interest.
Therefore we had to draw on other sources as well.
By the time of data retrieval, the diachronic Polish corpus PolDi encompassed 40 texts
ranging from the 14th to 19th centuries. This database was complemented by prose texts and
dramas from the electronic version of the Biblioteka zabytków Polskiego pi miennictwa
redniowiecznego (2006; www.staropolska.pl).20 Poetry was excluded, since our experience
with similar research in Russian and Croatian showed that the structures under investigation
usually do not occur in poetry, probably due to incompatibility of such lengthy expressions
with metrics.
To retrieve all inflectional (and orthographical) forms of the verbs mentioned above, we
searched the texts for the following expressions:
rzek.*
rzec.*
mówi.*
mowi.*
powie.*
wiedz.*
wyedz.*
20 We are aware of the fact that some of the texts from the Old Polish Period are offered there are modern
translations of the respective originals. Due to the bibliographic standards of the website, which understands
itself as a service for scholarly and educational purposes, these texts can be identified and excluded from the
corpus. Another issue to be addressed is normalization. All texts are to some extent normalized
orthographically, but since we are interested in syntactic issues, this fact is of minor consequence for us and
even reduces the complexity of queries, as no variants have to be considered.
25
Since macaronic literature is known to morpologically integrate Latin stems into Polish and
vice versa (cf. Keipert 1988), we also searched for Latin verba dicendi with the help of the
following expressions:
dic.*
dix.*
loqu.*
locut.*
ait
inqu.*
The results were then post-processed manually.
3.2. Historical development
3.2.1. Marking of direct and indirect speech

This is the first function attested. In early Slavonic texts in general, the distinction between
direct and reported speech cannot be drawn clearly for several reasons. Firstly, sentence and
clause boundaries cannot be identified clearly – a problem that concerns not only the
rendering of speech, but all syntactic issues. Secondly, reported speech is set off by the
characteristic that the speech content forms a subordinate clause which is argument of the
verbum dicendi in the superordinate clause and is consequently introduced by a
complementizer. However, this criterion is not yet obligatory in historical times (cf. Daiber
2009 for Russian Church Slavonic and Birzer 2012b for Old Church Slavonic; to the best of
our knowledge, no specific studies on older stages of Polish are available), which makes the
discrimination of direct and reported speech impossible if occurrence of the third
characteristic, namely shift of grammatical person (the probably most systematic description
of shift has been given by Večerka 2002, 416-423 for Old Church Slavonic), cannot be
determined clearly. This problem is especially notorious in texts with a non-involved observer
as narrator: all characters of the narration are then referred to in the third person, which makes
it impossible in many contexts to identify the participants of the original communicative
situation (compare the two examples from Modern Polish: Adam powiedział, że Anna
przyjdzie jutro. implies the two possible original communicative situations a) Adam
powiedziałŚ ―Przyjdziesz jutro.‖ where Anna is Adam‘s interlocutor, and b) Adam
powiedziałŚ ―Anna przyjdzie jutro.‖ where Anna is a third person absent in the original
communicative situation. Just as well, Adam powiedział, że przyjdzie jutro. denotes both (the
26
less proable) situation a) Adam powiedziałŚ ‖Przyjdzie jutro.‖ where Adam speaks about a
third, absent person, or b) Adam powiedziałŚ ‖Przyjdę jutro.‖ where Adam gives information
about himself). The vast majority of texts analyzed for this paper features a non-involved
observer as narrator and thus entails the problems described above. Therefore we treat direct
and reported speech together.
The marking of direct and indirect speech is effected with two Cxns. The first Cxn consists of
two verba dicendi; the first one relates the manner of content rendering and takes any
inflectional form, the second one is an (adverbial) participle in immediate precedence of the
speech content (42).
(42) … áby ∫ię wypełniło to co ie∫t rzeczono przez Proroká mowiącegoŚ Rozdźielili ∫obie
odźenia moie (Wuj NT Matth. 27/35)
‗… in order to become true what is said by the Prophet saying: They will divide my
garments among themselves.‗
In the second Cxn a noun from the semantic field of dicendi or cogitandi is accompanied by
the (adverbial) participle of a verbum dicendi (43-44).
(43) … usłyszeli głos z obłoku rzekący Ś Toć jest syn moj namilejszy.
‚… they heard a voice from the cloud sayingŚ this is my most beloved son.‗
(RozmPrzem)
(44) POL … oto głos z obłoku mowiącyŚ
DEM voice-NOM from cloud.GEN speak-PTCP.NOM.SG.M
Ten ieſt moy Syn miły.
DEM.NOM is my-NOM.SG.M son-NOM dear-NOM.SG.M
LAT … ecce vox de nube dicens
DEM voice-NOM from cloud-ABL speak- PTCP.NOM.SG
hic est filius meus dilectus.
this one is my dear son
‗… there was a voice from the cloud, sayingŚ This one is my dear son.‘
(WujNT Matth 17/5)
The Cxn with two verba dicendi continues to be used well into the 18th century (45-46), and
went through several developments of various impact on syntax.
27
(45) Przyjechał znowu ... Straszewski, listy oddał, ... i prosi mówiąc: Qui cito dat, bis dat.
‗Straszewski arrived again, handed over the letters and asks, speaking: Who gives fast
gives twice.‘
(Pasek. Pamiętniki.)
(46) Alexander Wielki słysząc z ust Filozofa, że wiele jest wiatów, płakał, że i jednego nie
zawojował, mówiąc: Heu me miserum, qui nec uno quidem potitus sum.
‗Alexander the żreat, hearing from the philosopher‘s mouth that there are many worlds,
cried that he had not conquered even one, speaking: Oh, what a poor lad I am, as I have
conquered not even one.‘
(Chmielowski. Nowe Atheny.)
The first development to be mentioned is the insertion of cataphoric expressions before or

after the verbum dicendi that refers to the content of the following utterance. Most commonly
used are deictic tak(o) ‗so‘ (50) – also an element of the modern KM aby / by / żeby tak
powiedzieć / rzec ‗so to say‘ – or NPs containing the demonstrative ten ‗this‘ (47-49). The
cataphoric expressions occur in both religious and worldly texts.
(47) … masz krolewstwo , tedy je ty krol . Miły Krystus mowiąc to słowo: Krolewstwo
moje nie jest [z] tego wiata.
‗ ‚You have a kingdom, thus you are a king.‘ Dear Christ speaking this wordŚ ‚My
kingdom is not of this world.‘‗
(RozmPrzem)
(48) Po tym tedy krótkim poswarku .... powiedział pan Lupa w te słowa:
‗After this short quarrel ... Mister Lupa said in these wordsŚ‘
(Górnicki. Dworzanin Polski.)
(49) Tu za pan Dersniak to powiedziałŚ Radbych ja wiedział, panie Wapowski, ...
‗And there said Mister Dersniak thisŚ I would like to know, Mister Wapowski, ...‘
(50) Pan Myszkowski tak na to powiedziałŚ ...‘
‗Mister Myszkowski answered (lit. spoke) on this soŚ ...‘
Let us now turn to developments that are all linked to the encoding of utterance content and
exert influence on syntax. As far as our data allows to reconstruct these developments, they
take place simultaneously; their results may be subsumed under the label (avoidance of)
syntactic hybridity.
28
The first syntactic development to be discussed, namely the AcI construction, is the result of
literacy contact with Latin (cf. Dubisz 2007: 9). Without doubt, (43) and (44) are two Polish
translations of the same Bible citation, which scrupulously follow the Latin syntactic pattern.
Translations of religious, especially biblical texts, and the pattern replications therein have
already received attention (cf. Moszyński 1994), as they play a special role in the discussion
about the exactness of syntactic replication. Since one aim of our paper is assessing the role of
language contact in the development of the KMs under investigation, it seems worthwhile to
compare our findings from religious translations with those from worldly ones.21 Since KMs
based on verba dicendi are the focus of this paper, we will also focus them in our analysis of
translation strategies. This, however, does not exclude the highlighting of other phenomena
that help to draw a fuller picture.
In the worldly texts we observe the coexistence of con- and divergence with the Latin
original, often within the same sentence. As a general rule, we can say that Latin uses the NcI
and (preferably) the AcI Cxn to set reported speech off from direct speech, and that the
Polish translations by and large follow that model.
The Latin original of (51) is characterized by a very complex sentence structure consisting of
a NcI with the matrix verb constat ‚it is clear‗ and an AcI with the matrix verb dictum esse ‗to
be said‘. Clear convergences with Latin that may not be explained away with similar Slavonic
syntactic structures are the following:
dictum esse is translated with the –no/-to form powiedziano ‗said‘, which at that time most
probably still expressed passive diathesis. In Modern Polish it is usually not accompanied by
the demoted agent. Here, the demoted agent is realized syntactically in the PP od niektórego
‗by somebody‘. As Szlifersztejnowa (1968: 133) and Meyer (2011: 242) note, variation
between the prepositions pzrez and od for the encoding of the demoted agent are attested in
older stages of Polish, whereas przez is the only acceptable variant in Modern Polish. Since
21 Unfortunately, it is rather difficult to identify texts, whose Latin original and Polish translation have both
 Collationes quas dicu[n]tur fecisse mutuo rex Salomon sapientissimus [et] Marcolphus facie deformis et
been preserved and which are both accessible. Therefore our corpus of parallel texts is confined to
turpissimus … a d the Polish t a slatio Roz owy, który iał król “alo o ądry z Mar hołte
grubym a sprosnym … Ja z Kosz zek ;
 Andrzej Wolan. De libertate politica sive civili (Kraków 1572) and the Polish translation O wol oś i
Rzeczypospolitej albo szlacheckiej “ta isła Du i go i z (Wilno 1606);
 Wespazjan Kochowski. Annalium Poloniae ab obitu Vladislai IV Climacter primus (1683) and the Polish
translation Roczników polskich od śmierci Władysława IV Klimaktery by Szymon Zabiełła (18th century)
We are aware of the fact that the analyzed translations stem from roughly two centuries and that differences in
the translation strategies thus might be induced by changes in translation practices or their philosophic
background. However, the analyzed Bible translations cover the same time span and are nonetheless uniform in
the applied translation strategies. Therefore, we may assume that variance in the translation strategies is not due
to the time factor, but rather the text genre.
29
both Polish od and Latin a also have the spatial meaning ‗from‘, it is rather possible that this
coincidence lead to od as second variant for denoting the demoted agent in Polish (cf. Rabus
2013: 283 for a similar semantic development of the preposition do as a result of Russian-
Polish language contact).
Case assignment in the subordinate sentence introduced by że is another issue of convergence:
Polish prefers the encoding of the predicative noun in the instrumental, but in (51) the
nominative is chosen, which results in a parallel to Latin where the subject and the predicative
noun also display the same case, although it is the accusative. The rendering of NcI and AcI
constitute clear cases of divergence: The NcI is not rendered at all, and the AcI is paralleled
by a complement clause, which is the regular means for realizing the semantic role ‗content‘
of verba dicendi in Slavonic.
(51) POL A tak prawdziwie jest od niektórego

CONJ so true-ADV be-3SG.PRS by somebody-GEN.SG
powiedziano, że barzo rzecz jest
say-PTCP.PASS.IMP COMP very thing.NOM.SG.F be.3SG.PRS
smaczna wolno ć ...
flavorful-NOM.SG.F freedom-NOM.SG
LAT [Verissime itaque dictum a
true-SUPERL.ADV therefore say-PTCP.PASS.NOM.SG.N by
quodam esse] NcI constat:
somebody-ABL.SG be-INF be_clear-3SG.PRS
[dulce nimirum bonum esse libertatem]AcI …
sweet-NOM.SG.N indeed good-ACC.SG be-INF freedom-ACC.SG
‗Thus, clearly it has been said most rightfully that freedom is a very sweet good.‘
(Wolan, Andrzej. De libertate politica sive civili)
Example (52), however, replicates the Latin AcI for the encoding of the semantic role
‗content‘ in two cases. At first glance, the replications are not that obvious, as both nazywać
‗call‘ and rozumieć ‗understand‘ govern the accusative. At second glance, the two renderings
differ structurally, as the second one with rozumieć ‗understand‘ as matrix verbs clearly
replicates the Latin structure with noun and predicative adjective in the accusative. The first
one with nazywać ‗call‘ as matrix verb is a hybrid between Latin and Polish syntactic
30
structures. The predicative adjective is in the instrumental, as secondary predicates in Polish
usually are, but the copula is in the infinitive, as required for the Latin AcI.
Furthermore, the chosen equivalents for the Latin verba dicendi also deserve mentioning.
Nazywać ‗call‘ for dicere ‗say, speak‘ is not a literal translation, but results in rather idiomatic
Polish. The possible reasons for translating of present tense inquit ‗he is saying‘ with preterite
rzekł ‗he said‘ will not be discussed here, as this might be a deliberate decision grounded in
narrative effects.
(52) POL I dlatego, gdy Kallistena filozofa przeto

and because when Kallisten-NOM philosopher-NOM therefore
szczę liwym niektórzy być nazywali, iż
happy-INSTR.SG some-NOM.PL be-INF call-PST.3PL that
u Aleksandra barzo hojnie i rozkosznie żył,
he lived at Alexander‘s place lavishly and luxuriously,
"i owszem, ja – rzekł – biednego i
and yes I-NOM say-PST.3SG poor-ACC.SG.M and
nieszczę liwego być rozumię ...
unhappy-ACC.SG.M be-INF understand-PRS.1SG
LAT Ac proinde cum [Callisthenem philosophum
and therefore when Callistenis-ACC.SG philosopher-ACC.SG
hoc nomine beatum]AcI quidam
this.ABL pretext-ABL happy-ACC.SG.M some-NOM.PL
dicerent, quod
say-CONJ.PST.3PL because
apud Alexandrum lautissime ac splendide viveret,
he lives at Alexander‘s place most respectably and lavishly,
Immo ego, inquit, [miserum et

in_contrast I-NOM say-PRS.3SG miserable-ACC.SG.M and
infelicem]AcI puto ...
unhappy-ACC.SG consider-PRS.1SG
‗And therefore when some call the philosopher Callisthenes happy under the pretext that
he lives at Alexander‘s place most respectably and lavishly, I, in contrast, consider him
miserable and unhappy.‘
31
(Wolan, Andrzej. De libertate politica sive civili.)
In contrast, example (53) shows strict adherence to the formal composition of the Latin
original:
(53) Marchołt odpowiedział i rzekłŚ…
Marchołt-NOM answer-PST.3SG and say-PST.3SG
Marcolphus rˉndit [respondit] 7 ait: …
Marcolphus-NOM answer-PST.3SG and say-PST.3SG
‗Marchołt answered and saidŚ ...‘
(Jan z Koszyczek. Rozmowy które miał król Salomon mądry z Marchołtem grubym a
sprosnym ...)
To summarize, the analyzed translations show Latin influence predominantly on the syntactic
level, especially the marking of indirect speech, and this influence also affects how the
arguments of verba dicendi are realized syntactically. Regarding lexical choices, idiomaticity
seems to prevail over adherence to the Latin model.
Another important contact language for Polish, namely German, also features the AcI
construction, but, in contrast to Latin, it is restricted to verbs of perception. In our corpus data
of original Polish texts, AcI constructions are attested for both verbs of perception (54) and
verba dicendi (55-56). This fact, and the more so macaronic clauses where the AcI is built
with Latin lexical material (55) can be considered evidence that Latin influence lead to AcI
constructions in Polish. It cannot be excluded, however, that German fostered this (temporary)
development.
Note also the syntactic hybridity of the examples: the word order in (54) resembles the
organization of a Polish hypotactic sentence that encodes the the semantic role ‗content‘ of
the verbum dicendi as subordinate clause, although we are dealing with an AcI whose matrix
verb usually takes clause-final position. In (55) the typical word order of the AcI Cxn is
retained, the fact that dobrą ‗good‘ can be interpreted as both accusative and instrumental
makes the syntactic structure oscillate between AcI and secondary predication, where the
Polish predicative adjective takes the instrumental.
At the same time, the AcI helps to resolve the aforementioned syntactic hybridity that usually
arises in speech act rendering contexts where the narrator and the subject of the rendered
situation are both in the third person, as the AcI marks reported speech.
(54) Gdy widziała Alexandra [być w tak wielkim
32
when see-PST.3SG.F Alexandra.NOM be-INF in so huge-LOG.SG
niebespieczeństwie dzieci swoje]AcI ...
danger-LOC.SG child-ACC.PL POSS.REFL.ACC.PL
‗When Alexandra saw that her children were in such huge danger …‘
(55) ... przeto Orpheus powiedział, [Jovem być
therefore Orpheus-NOM say-PST.3SG.M Jupiter-ACC be-INF
masculum et feminam]AcI.
male-ACC and female-ACC
‗... therefore Orpheus said that Jupiter was both male and female.‘
(56) Niechaj pirwej tym panom odpowiem,
let at_first that-DAT.PL gentleman-PL.DAT answer-FUT.1SG
którzy [piękno ć nie zawdy dobrą być]AcI powiedają ...
who-NOM.PL beauty-ACC.SGnot always good-ACC.SG be-INF say-PRS.3PL
‗Let me at first answer those gentlemen who claim that beaty is not always good ...‘
If the AcI may be considered a means to mark indirect speech, then the process of mówiąc
‗speaking‘ loosing the status as secondary predication may be described as the development
of a direct speech marker.
The first step in this direction is illustrated by example (47), where the sentence containing
the AP mówiąc ‗speaking‘ features miły Krystus ‗dear Christ‘ as canonical subject in the
nominative, but lacks a matrix verb. In other words, the AP still has a point of coreference,
but cannot be classified any more as secondary predication, as the sentence lacks a primary
predication.
Finally, (57) illustrates the complete loss of subject co-reference: mówiąc ‗speaking‘
constitutes a sentence of its own with no NP, not to speak of a matrix verb that might serve as
point of co-reference. Furthermore, the context makes clear that the covert subject of the
adverbial participle refers to niektorych ‗some [Lutherans]‘, i.e. the object of the preceding
sentence. Thus we may state that in contexts of this kind mówiąc may be regarded as a speech
marker, but no more as an adverbial participle in sensu stricto.
33
(57) Papież po∫łał do Niemiec trzech Bi∫kupow z Indultem / áby namawiáli Luteryany
kapłany ná Rzym∫ką wiárę / […] ále niekthorych-ACCj ná tho niemogli przywie ć /
mowiącjŚ lepiey ie∫t Bogá ∫łucháć niż ludzi.
‗The Pope sent to Germany three bishops with an indult / that they persuade the Lutherans
of the Roman Rite / but somej they could not persuade of this / speakingj: it is better to
obey żod than humans.‘
(BielKron 234, 30v, 41)
Another issue that needs to be addressed here is the semantic erosion of verba dicendi in
discourse structuring function, although this is best illustrated by mówiąc as quotation marker
(58-59), i.e. a function that develops out of speech marking, but may be considered the
precursor of the modern quotative KMs.
The semantics of the verb mówić ‗speak‘ with the government pattern X mówi Y / o Y-u can be
rendered with the following preliminary explicationŚ ‗with the help of their voice apparatus, X
produces sounds conveying content Y‘. The meaning of the speech and citation marker
mówiąc (government pattern mówiącx Y) can be described as followsŚ ―X produces language-
based content Y (sprachlichen Inhalt Y)‘. Thus, the development of the marker mówiąc goes
along with the loss of the semantic component ‗mode of articulation‘.
(58) … [w] dzisiejsze ewanjelije było pisano rzekąc tako , iże wyszło jest przykazanie było
‗… in today‘s gospel was written speaking so, that an order had been issued …‘
(KazGn)
(59) O RZECE SABBATICUS, albo SABBATIUS, od sabaszu żydowskiego rzeczonej, pisze
Starożytny Author Pliniusz lib. 31, cap. 2, mówiący: in Judaea rivus Sabbatis omnibus
siccatur.
‗About the word sabbaticus, or sabbatius, derived (lit. said) from the Jewish Sabbath, the
ancient author Plinius, lib 31, cap.2, writes speaking: in Judaea rivus Sabbatis omnibus
siccatur.‘
(Chmielowski. Nowe Ateny...)
In connection with the development of the speech marker mówiąc we would like to point out
a particular case of syntactic hybridity in the rendering of speech (60-61): although the
pronoun shift leaves no doubt that we are dealing with direct speech, both the marker mówiąc
and the complementizer że ‗that‘, which usually introduces reported speech, are used. Since
34
we were confronted with this phenomenon only in Pasek‘s work, it is most likely an
idiosyncracy of this author, but it underlines the striving to set the protagonists‘ utterances off
from the narration as such.
(60) ... i tym gloriabantur mówiąc to, że "my w to wierzymy, co i wy, daremnie nas
nazywacie odszczepieńcami".
‗ ... and they boasted, saying this, that „we believe in the same as you do, in vain you call
us apostates.‖‘
(61) ... król ucieszył się nadzieją mówiąc, że "mnie pan Pasek dawno znajomyś wiem, że mi
jej nie odmówi."
‗ … the king was pleased by the hope, saying that ―I have been acquainted with Mister
Pasek for a long timeś I know that he will not refuse me this.‘
Another method for distinguishing the narration from the utterances embedded therein is the
juxtaposition of Polish and Latin. To the best of our knowledge, studies on the linguistic
aspects of macaronism are rather rare and have been concerned mainly with morphological
issues. Thus, based on the literary work of Orzelski, Keipert (1988) gives an account of the
morphological means for Latinizing the inflection of Polish lexical material, with focus on the
declension of nouns, the declension and gradation of adjectives and adverbs, and the
conjugation of verbs. This means that, so far, the focus has been on the mutual morphological
integration of lexical items but not on the distribution of Polish and Latin lexical items and the
functional aspects thereof. As our examples show, the juxtaposition of Polish and Latin is
used in two types of speech marking: the first and probably most natural case are citational
contexts, where quotes from Latin sources are rendered in the original (45-46; 59; 62).
Instances like (55) are of another type, as we are not dealing with a verbatim citation, but a
reported speech in the shape of an AcI construction that is set off from the general narration
by using Latin. Thus, in the terminology of language contact studies, we may describe this
phenomenon an instance of functional code-switch (cf. Riel 2004: 23).
(62) … jaki Heretyk … wolałby był być Jeremiaszem, mówiącym: A, A, A, Domino DEUS
ecce nescio loqui, w takiej materyi którą posuit Pater Celestis in sua dispositione.
35
‗... some heretic … would have preferred to be Jeremias, who speaks (lit. speaking) A, A,
A, Domino DEUS ecce nescio loqui, in the material form which Pater Celestis gave in his
disposition.‘
To summarize, the development of KMs based on verba dicendi begins with a Cxn for the
marking of direct and reported speech. It consists of two verba dicendi, of which one is
semantically more complex and renders the mode of speaking, whereas the second verbum
dicendi is characterized by its infinite morphology, usually in the form of the (adverbial)
participle. This Cxn is ambigious in the sense that it does not allow to distinguish between
direct and reported speech, if the protagonists of the narration and the situation rendered via
speech are both in the third person, or if they coincide. Two strategies to disambiguate direct
and reported speech evolve. The first one, namely the usage of the AcI Cxn for the encoding
of reported speech, is most probably due to literacy contact with Latin. Secondly, through the
loss of syntactic co-reference and the erosion of the semantic component ‗mode of
articulation‘, the adverbial participles mówiąc ‗speaking‘ and rzekąc ‗saying‘ turn into a
markers of direct speech and quotations.
Furthermore, the functional code-switch to Latin in order to set off utterances – be they in the
form of verbatim quotations or reported speech – is an additional means that – just like the
AcI Cxn – has been used mainly in the 17th and 18th century when macaronism was the
preferred literary style. Thus, it seems that an extralinguistic factor, namely a literary style,
exerted influence on the syntax of Polish.
In one way or the other, all functions and Cxns described below take Cxns used for speech
marking as point of departure. The data at our hand varies in the sense that it does not allow to
completely reconstruct the historical development of each function and Cxn. Therefore, we
will set out with the stance-marking function of the Cxn based on mówiąc, as these data are
most comprehensive and the Cxn serves as point of departure for other functions.
3.2.2. The stance-marking function of the Cxn based on the adverbial participle
At the turn of the 15th to the 16th century we witness the cooccurence of Cxns based on finite
verb forms and the AP of verba dicendi respectively. These Cxns are modified not by
cataphoric elements, but adverbs (63-65; 67) and, to a much lesser extent, NPs (66). They all
classify text segments according to metatextual information, namely their length or
interdependence with other text segments. For two reasons, this also implies the expression of
36
speaker‘s stance: the classification of a text segment as long or short bases on a personal
judgement, if – as in (63) – no tertium comparationis is available. Objective length is the one
thing, content the other. So even if a tertium comparationis for classifying length is available
(65), one text segment can be considered a shortened version of a preceding text segment only
if they render the same content. Whether contents coincide is, however, subject to personal
judgement and thus expresses speaker‘s stance.
Quite interestingly, the Cxns with finite verb forms and the AP respectively differ in their
scope and valency structure. The Cxn with finite verb forms is monovalent and scopes over
one constituent of its host utterance (63-64), whereas the Cxn based on the AP is bivalent and
scopes over two text segments of different size (65-68). A possible explanation for this
difference will be discussed below.
(63) To, com ja tu krótko powiedział, snadź by już pana [...] uczynić dobrego mogło ...
‗This, what I have already said, apparently could already have made you a good man … ‗
(Gornicki. Dworzanin polski.)
(64) Kiedy tu pan Wapowski przestał, jakoby, rzeczy swej dokończywszy, dalej
mówić niemiał, powiedział pan BojanowskiŚ Inaczej tego żaden tu rzec nie może, panie
Wapowski miły, ....
‗When Mister Wapowski stopped here, as if, having finished his issue, he had nothing
more to say, Mister Bojanowski saidŚ ‗Otherwise than that no one here can say it, dear
Mister Wapowski, …‘
(Górnicki. Dworzanin polski.)
(65) Lecż te inne cżtery [punkty wyznania wiary]/ wolnie á vprzemymie Øj wyználi-PST.3PLj
/ iż tak ieſt / á inacżey mowiący-NOM.PLj / przeklęći ſą.
‗But the other four ones [principles oft he confession of faith] / they confessed freely and
friendly / that it is so / and in other words (lit. otherwise speaking) / that they are damned.‘
(SkarJedn 275)
(66) Bo mowiąc ku prawdzie , kto taką rzecz słyszał Ś Cztyrzy dni leżał umarły w grobie
a on ji skrzesił?
‗Because speaking the truth, who has heard such a thing: For four days he had been lying
dead in the grave and he has resurrected?‘
(Rozmy lanie Przemyskie)
37
(67) ... dawszy sie pirwej długo prosić, nakoniec przyzwoliła. I kazała to Sinorixowi
powiedzieć, który z wielką rado cią począł sie hnet starać, aby ta rzecz rychły skutek swój
wzięła. Owa krótko powiedając przyszedł czas lubu …
‗ ... after letting him beg for a long time at first, she finally agreed. And she ordered to tell
this Sinorix, who immediately and with great pleasure started efforts, so that this
endeavour take its fast course. So, in short (lit. shortly speaking), the time of the wedding
approached …‘
(68) ... oboje tego pragnęło, aby takowa miło ć szczę liwy koniec wzięła, to jest, aby s sobą
w małżeństwo przyszliś tegoż im też wszytcy w onym mie cie ludzie życzyli, okrom jej ojca,
który dziewkę swą komu bogatszemu ... dać wolał. W czym uboga dziewka okrutnemu
ojcu sprzeciwić sie nie miała [...] Krótko powiedając doszło to tam nieszczęsne
małżeństwo ...
‗... both wished themselves, that this love take a happy end, that is, that they marry each
other; all people in that place also wished them this, except her father, who wanted to
marry his girl to a rich man. The poor girl did not dare to oppose the ferocious father … In
short (lit. shortly speaking) it came to this infelicitous marriage …‘
As with the speech marker described above, the development of the stance marker also goes
along with the gradual dissolution of co-reference between the subjects of adverbial participle
and the matrix verb, and with the gradual loss of the AP‘s status as secondary predication.
This process must have taken place between the 15th and 16th century, as our examples show.
In (65) we are dealing with a present participle (out of which the adverbial participle
developed) that agrees in case and number with the (elliptic) subject of the matrix clause.
Thus, co-reference of subjects is still maintained, although the function of inaczej mówiący is
that of its modern equivalent inaczej mówiąc ‗in other words (lit. in another way speaking)‘ –
it connects two text segments and marks that the latter one is a reformulation of the former
one‘s content. źxample (66) represents the next step in the dissolution of subject co-reference,
as it allows two readings. If we assume that subject co-reference is still maintained and the AP
clause bears a conditional reading, the whole sentence can be interpreted as ‗Who, if he
speaks the truth, has heard such a thingŚ …‘. However, if we assume that subject co-reference
applies no more and that the speaker is the covert subject of the AP, the only possible reading
is ‗Honestly speaking (lit. speaking the truth), who has heard such a thingŚ …‘. (67-68) are
38
clear examples for the speaker as covert subject, because mówić requires an animate first
argument, whereas the subjects of the matrix clauses – lub ‗wedding‘ and małżeństwo
‗marriage‘ – are inanimate abstracts.
In our data for the 15th and 16th century the stance-marking Cxn based on mówiąc ‗speaking‘
was represented by three types: krótko mówiąc ‗in short (lit. shortly speaking)‘, inaczej
mówiąc ‗in other words (lit. in another way speaking) and ku prawdzie mówiąc ‗speaking the
truth‘. All three convey speaker‘s stanceś additionally, the two former ones function as
reformulating connectives.22 Connectivity implies two argument slots. Although verba
dicendi are ascribed (at least) two argument slots (namely the agentive speaker X and the
content Y as in X mówi Y / o Y-u ‗X speaks Y / about Y‘), this cannot be the reason for the
bivalency of the discussed reformulating connectives: the speaker as covert subject of the
adverbial participle mówiąc has already taken slot X. Rather, the two argument slots of the
reformulating connective seem to result from a blending of syntactic and semantic argument
structures: the adverbial participle as syntactic structure opens an argument slot for its matrix
clause, and the semantics of the verbum dicendi mówić, which takes on the form of the
adverbial participle, opens the semantically founded argument slot Y for content.
In comparison to the three types attested in the 15th and 16th century, we witness a
considerable rise of type frequency in the 18th century. In order to better understand the
development in the 18th century, some words are in order on the definition of type. We can
define types on formal grounds, i.e. their orthographic and morphological form, or on
semantic (and thus functional) grounds.
In the 18th century, we count ten formal types, but only seven semantic types (cf. Table 1),
which means that there must exist some synoynyms that differ formally. In the case of
‗honestly speaking‘, we are confronted with a morphological difference, as the two synonyms
base on two different inflectional forms of the same base word. The picture for ‗exactly
speaking‘ is another one, since the stems of the two bases – wła ciw- and partykularn- – bear
the Polish adverb ending –ie (partykularnie also the stem-building suffix –n-), but the third
base – in particulari – is Latin in stem, orthography and morphology and must be the source
for the orthographically and morphologically integrated Polish loan word partykularnie. If we
have a closer look at the types, two more bear similar characteristics: generalnie is derived
from the Latin root general- with the help of Polish suffixes, and genuine is Latin in both
22 These findings verify the hypothesis put forward in (Birzer 2015), where, based on semantic bridges and the
distribution of functions across spoken and written Modern Polish, stance marking in general and the
reformulating connective in particular have been identified as the point of departure for the development of all
other functions fulfilled by KMs.
39
stem and suffix morphology (which is, however, orthographically very close to the Polish
ending –ie). Since four out of ten types are clearly influenced by Latin, we need to address the
question which overall role Latin played in the development of stance markers based on the
adverbial participle.
Table 1. Formal and semantic types of stance markers based on the adverbial participle
in the 18th century23
formal types semantic types
inaczej mówiąc ‗in other words (lit. in another way
speaking)‘
krótko mówiąc ‗in short (lit. shortly speaking)‘
po prawdzie mówiąc ‗honestly speaking (lit. the truth speaking)‘
prawdę mówiąc
rzetelniej mówiąc ‗besser gesagt‗ ‗in a better formulation (lit. better speaking)‘
wła ciwie mówiąc
partykularnie mówiąc ‗exactly speaking‘
mówiąc in particulari
generalnie mówiąc ‗generally speaking‘
genuine mówiąc ‗straightly speaking‘
Given that the same construction based on an adverbial participle can also be found in other
Slavonic languages (cf. Birzer 2012 for Russian and Birzer (accepted b) for Croatian), we can
exclude that the general syntactic pattern for this construction has been replicated from Latin.
Nonetheless, we found a Latin stance marking parenthesis in our corpus (69), which is quite
remarkable, as the ten stance markers from Table 1 produce an overall token frequency of just
16. The Latin parenthesis bases on the optative of dicere ‗say‘ in the first person singular, i.e.
it is clearly produced by the speaker and not by a protagonist of the narration. The verbum
dicendi is modified by an adverb, which is prototypical for the stance marking Cxn. Due to
this structural parallel it is likely that the array of Latin modifiers that was transferred to
Polish served as resource for the growing type frequency of the Polish Cxn. Our corpus data
for the types with the meaning ‗exactly speaking‘ backs this hypothesis: in Chmielowski‘s
Nowe Ateny… , published in 1745, we find both the Latin source word in its Latin
23 for reasons of space we use mó ią as pla eholde fo all th ee attested APs ówią , rzeką and
powiedziawszy.
40
morphological shape (70), as well as the orthographical and morphological adaptation of its
stem into Polish (71). Four years later, Kitowicz uses in his Opis obyczajów… the same
semantic type but with an originally Polish adverb as modifier (72).
(69) .... wszystka kompanija – verius-ADV dicam-CONJ.PRS.1SG – pouciekała ...

‗... the whole company – I shall tell most truefully – deserted …‘
(70) A dopieroż tyle było Atheńskich, Lacedemońskich, Rzymskich […] Rzeczypospolitych,
które […] żyły wolno cią. A jak tylko umbra, a bardziej larwa niewoli w Senat Rzymski
(mówiąc in particulari zazierać poczęła, ...
‗And at first there were so many Athenian, Lacedemonian and Roman republics, which
lived in freedom. But when only the shadow, or rather the ghost of illiberty in the Roman
Senate (began precisely speaking to lurk …‘
(71) Partykularnie mówiąc o Wolno ci Polskiego Narodu, klejnotem takim ją nazywam,
któremu żaden Jubiler nie znajdzie szacunku.
‚Precisely speaking about the freedom oft he Polish people, I call it a treasure whose value
no jeweler can estimate.‗
(72) W osobie albo wła ciwie mówiąc w wizerunku osoby … wyrznięta była dziura okrągła
... tak wielka jak hostia ...
‗In the person or, more precisely, in the picture of the person … a round hole the size of a
host was cut out …‘
(Kitowicz. Opis obyczajów...)
This leads us to the hypothesis that the expansion of the prototypical stance marking Cxn ADV
+ AP is the result of the intricate interplay between catalyzation and matter replication (in
Sakel‘s (2007) sense). Most probably, the structural and functional parallels between the Latin
parenthetical Cxn and the Polish ADV + AP Cxn lead to matter replication from Latin, which
resulted in rising type frequency. Therefore, the Latin parenthetical Cxn as such was the
catalyzer for the expansion of the Polish ADV + AP Cxn, and the concrete types of the Latin
Cxn served as resource for mattern replication.
Our corpus material from the 18th century includes two more instances of matter replication,
namely generalnie mówiąc ‗generally speaking‘ and genuine mówiąc ‗straightly speaking‘.
41
Unfortunately, our 18th century data does not comprise any synonyms to these types, and the
structure of NKJP does not allow to conduct a sensible real-time study of the developments in
the 19th century for the following reasons:
The PELCRA interface of the NKJP offers the konkordancje and the kolokator function; both
functions can be applied to the so-called balanced subcorpus, which contains texts generated
after 1945, and to the full corpus, whose contents are not described on the NKJP homepage.
The konkordancje function allows the user to define their own subcorpora by time spans24,
and offers a wpm-frequency chart for the searched item via the czas function. Unfortunately,
the czas function is available only for texts from the years 1988 to 2010 and works only with
single words, not collocations. The kolokator function for research on collocations works only
with the balanced corpus, which makes it non-utilizable for research on data from the 19th
century.
Therefore we opted for a ―retrospective‖ approach, i.e. we identified all Cxns of the type ADV
+ mówiąc / powiedziawszy with the help of the query [orth=",|.|-"] [pos=adv]
[orth="(mówiąc|powiedziawszy)"] [orth=",|.|-"], identified groups of synonymous types and
filtered out groups where at least one type features an adverb with Latin root (Table 2). By
this method we identified twelve relevant groups. For some of them, namely kolokwialnie,
cynicznie, elegancko, metaforycznie, konkretnie, precyzyjnie and technicznie, it is rather likely
that they are not direct Latin loans, but came into Polish via contact with other languages.
Nonetheless, they all show that the stance marking Cxn expanded with the help of matter
replications that were then doubled by types with an originally Polish adverb. The trywialnie /
banalnie group is the only exception to this rule in the sense that banalnie is also a loanwoard,
yet from French. The ‗precisely / exactly‘ group is an interesting continuation of the group
described in examples (70-72), as the original matter replication in particulari / partykularnie
did not survive into Contemporary Polish, but the semantic pattern was maintained with
wła ciwie(j) as continuous representative that was backed by later synonymous types that
were the result of new matter replications.
Table 2. Groups of synonymous Cxn types with at least one Latin adverbial root
token frequency of
adverb semantics Latin source word Cxn type
kolokwialnie colloquially colloquium 21
24 our query for ówią in texts from the full corpus written between 1800-1900 gave some matches, so we
th
may conclude that the full corpus contains some texts from the 19 century. Another drawback of the NKJP is
the fact that no information on the corpus size – be the respective corpus predefined or personalized – is given,
which makes wpm counts and thus the comparison of subcorpora covering different time spans impossible.
42
cynicznie cynically cynicus 8
elegancko elegantly elegans 6
generalnie generally generalis 14
(naj)ogólnie(j) generally 160
metaforycznie metaphorically metaphora 10
obrazowo metaphorically 44
(naj)delikatnie(j) mildly delicatus 194
(naj)łagodnie(j) mildly 30
popularnie popularly popularis 5
po_ludzku popularly 32
konkretnie(j) precisely / exactly PPP concretus 15
precyzyjnie(j) precisely / exactly PPP praecisum 15
dokładnie(j) precisely / exactly 76
wła ciwie(j) precisely / exactly 18
brutalnie(j) roughly brutalis 27
serio seriously serius 26
poważnie(j) seriously 66
technicznie technically technicus 1
trywialnie trivially trivialis 1
banalnie trivially 1
We would like to close our observations on the role of language contact in the development of
the stance-marking Cxn with some considerations on why we deem Latin as both catalyzer
and source language for the matter replications, although Polish was also in long-lasting
contact with other languages, foremost Czech, German, Ruthenian and Russian (for a
comprehensive survey of all contact situations with a special focus on the those within in the
Slavonic language family cf. Rabus 2013: 202-316)? This question is the more legitimate, as
three of the aforementioned languages are Slavonic ones, and both Czech and Russian are also
known to have a stance marking Cxn based on the adverbial participle.
German features a semiproductive stance marking Cxn based on the passive participle gesagt
‗said‘ with the syntactic structure ADV + PTCP.PASS, i.e. the degree of structural similarity
between the Polish and the German Cxn is roughly the same as between the Latin and the
Polish ones. Yet our historical corpus data does not display traces of German influence in this
realm, i.e. we have not found stance marking Cxns based on the passive participles mówion-,
powiedzian- or rzeczon- and we have not found any instances of matter replication. This
makes it rather implausible that German exerted influence on the development of the stance
marking Cxn.25
25 even if there was some German influence which we could not trace due to the corpus size, No o iejski s
(2007) findings might explain why the results of this influence have not made their way into Contemporary
43
Czech exerted considerable influence on Polish up to the 16th century and was itself in
intensive contact with German. Just like Croatian, which was also part of the Austrian
Empire, Czech possesses two competing stance marking Cxns – one based on the adverbial
participle and the other one based on the passive participle, most probably the result of
language contact with German. Since the Polish stance marking Cxn started to develop only
very slowly in the 16th century, i.e. in a time when Czech influence on Polish was already on
the verge of decreasing, it is rather unlikely that Czech influence catalyzed the expansion of
the Polish Cxn, the more so, as the structurally similar Czech Cxn was itself competing with
the Cxn ADV + PTCP.PASS replicated from German.
Just like in Polish, the ADV + AP Cxn is the semiproductive means to form stance markers in
Contemporary Russian. This fact makes the role of East Slavonic varieties as catalyzer for the
spread of its Polish equivalent rather probable. To check this assumption we drew on data
from Karłowicz (1984) and Kara (1996).
Karłowicz (1984) is the publication of the dictionary manuscript Podręcznik czystej
polszczyzny dla Litwinow i Petersburszczan that must have been compiled in the late 1870ies
or very early 1880ies (cf. Kaupu & Smułkowa 1984: 87), i.e. in a time when KMs based on
verba dicendi were growing rapidly in number. Since Matras (1998) has shown that discourse
structuring elements are affected by language contact rather early, it is not too surprising that
Karłowicz features three conjunctions (między tem ‗meanwhile‘ (1984Ś 50)ś tak jak
‗because‘(1984Ś 67) tem nie mniej ‗nonetheless‘ (1984: 68) and two KMs (po drugie
‗secondly‘(1984Ś 57)ś po pierwsza rzecz …. po druga rzecz ‗firstly ... secondly‘ (1984Ś 57)
whose source is Russian.
Kara (1996) is concerned with Russicisms in the Polish of the Partition period. Kara ‘s major
merit lies in the fact that – although the monograph title speaks of rusycyzmy słownikowe
‗lexical Russicisms‘ – she also discusses structural borrowings and, based on the analysis of a
press corpus, establishes their first occurrence. Just like Karłowicz (1984), Kara identifies a
whole number of conjunctions and KMs (partially the same as Karłowicz) of Russian origin
(cf. Kara 1996: 314-315), but also prepositions based on adverbial participles, such as nie
Polish. Nowowiejski (2007: 17) states that especially in the Austrian Partition of Poland German exerted not
only lexical influence, but influenced also the stylistic-syntactic level. Among others pattern replications to place
that lead to phraseological expressions belonging to the official register. Nowowiejski highlights that quite a lot
of pattern replications survived puristic endeavours in the second half of the 19th century, as they were
considered a ―minor evil‖ and were used as a regular means of replenishing the lexicon (2007Ś 17). Żrom this we
may draw the conclusion that KMs based on the past participle – if such had ever existed in Polish – were
recognized as Germanisms and became subject to purification, the more so, as Polish displayed no lexical ―gap‖
due to the existence of the parallel construction based on the adverbial participle.
44
bacząc na ‗regardless of‘ś nie patrząc na ‗regardless of‘ś nie zważając na ‗regardless of‘
(1996: 313), yet no KMs based on verba dicendi. The mentioned prepositions came into
Polish between 1864-1905 (cf. Kara 1996: 324), i.e. at a time when KMs based on verba
dicendi were already fairly well rooted in the language and Russian had already exerted
influence on Polish for more than half a century.
In both Karłowicz (1984) and Kara (1996) no KMs based on verba dicendi are listed. We
may regard this fact as evidence that no matter replication took place. This finding is
confirmed by the matches of our aforementioned NKJP query [orth=",|.|-"] [pos=adv]
[orth="(mówiąc|powiedziawszy)"] [orth=",|.|-"] that produced no lexical elements that were
obviously borrowed from East Slavonic varieties. However, on the given empirical basis we
cannot exclude that East Slavonic varieties had some catalyzing effect on the Polish ADV + AP
Cxn.
Finally, some words are in order on the adverbial participles powiedziawszy ‗having spoken‘
and rzekąc ‗speaking‘, as they are potential competitors of mówiąc as base for the semi-
productive ADV + AP Cxn. Rzekąc is attested as speech marker until the 16th century (43; 58),
but not as base for the stance marking Cxn. Due to the size of our diachronic corpus, we
cannot exclude that its occurrence in this function is simply not mirrored in the corpus.
Therefore, two scenarios are possible. Since the stance marking function developed out of the
speech marker, it might be that rzekąc was also used as base for the stance marking function,
but was supplanted by the productive pattern based on mówiąc. This development could be
described as a tightening of the paradigm in the sense that the number of adverbial participle
forms that serve as basis for the stance marking Cxn decreases. On the other hand, it is also
possible that rzekąc never served as base for the stance marking Cxn, but specialized as
speech marker and consequently got out of use when this function was taken over by other
structural means.
The case of powiedziawszy – the perfective adverbial participle – is different, as it has not
been attested as a speech marker, but it occurs in contexts where it marks the end of an
utterance, which is anaphorically referred to with to ‗this‘ (73-74). So the syntactic structure
and thus the possibility to take on the stance marking function is the same as for the speech
marker, only the position in the text differs.
(73) A to powiedziawszy wstał i onej rozmowie uczynił koniec.

‗And having said this, he got up and made an end to this conversation.‘
45
(74) Powiedziawszy to pan Wapowski, umilkł.
‗Having said this, Mister Wapowski became silent.‘
Yet in contrast to rzekąc, powiedziawszy is attested as base for the stance marking Cxn,
although only with few types that are furthermore doubled by types which are based on
mówiąc and have a much higher token frequency (cf. Table 3).26 Quite interestingly, the
number of types based on powiedziawszy is not only low, but the types also differ
significantly as to token frequency.
Table 3. Token frequencies for powiedziawszy-based types in comparison to mówiąc-

based ones
modifier powiedziawszy mówiąc
prawdę ‗truly (lit. the truth)‘ 314 712
szczerze ‗frankly‘ 84 845
ci le(j) ‗(more) strictly‘ 5 223
inaczej ‗otherwise‘ 3 666
prosto / najpro ciej ‗plainly / most plainly‘ 2 66
dokładniej ‗more precisely‘ 1 56
lepiej ‗better‘ 1 3
This allows to construct two possible historical scenarios. In the first one, powiedziawszy has
served as base for the stance marking Cxn also at different historical stages of Polish (but is
not attested due to corpus size) but for some reason became inproductive. Consequently, the
remaining types in Contemporary Polish are to be considered (petrified) lexicalized items, and
the fate of powiedziawszy, just like that of rzekąc, is part of the tightening of the paradigm.
This explanation would be very plausible if it were not for the five types with a token
frequency of five and below. The petrification of highly frequent items is not uncommon –
note that the type with prawdę / po prawdzie is one of the first attested types and that szczerze
‗frankly‘ stems from the same semantic field – but there is no obvious explanation for why
the types with low token frequency should have petrified. On the other hand, they can be
explained away with occasional analogies.
26 the used corpus query was [orth=",|.|-"] [] [orth=powiedziawszy] [orth=",|.|-" orth!=że

46
Analogy is also the driving force in the second scenario. Here, the types based on
powiedziano are built in analogy to those based on mówiąc only at a rather recent stage in the
history of Polish, which would also explain occasionalisms.27
However, a final answer to this question can only be given once a much more comprehensive
diachronic corpus will be available.
3.2.3. Development of the contextualizing Cxn based on mówiąc ‘speaking’

In the 16th century we find several examples that allow to reconstruct the emergence of
contextualizing parentheses, yet with finite verb forms. In the three examples to be discussed
(75-77), the clause containing the verbum dicendi identifies the source of knowledge for the
facts related in other parts of the sentence, but its degree of syntactic integration varies. It is
highest in example (75), where the relationship between point of information reference (złoty
wiek ‗golden age‘), the source of information and the information itself is organized in a
complex hypotactic construction. The point of information reference forms the head noun for
a relative clause indicating the source of information; the information itself is encoded in a
complement clause dependent on the verbum dicendi powiedać ‗speak, say‘, which forms the
predicate of the relative clause. This syntactic structure clearly separates information
reflecting the speaker‘s own line of thought from related information. Casually speaking, one
could say that the indication of the information source has scope only over the content of the
complement clause.
Example (76) is ambiguous as the dependence of the complement clause introduced by iż
‗that‘ is unclear. In the case of interpretation a., the complement clause has to be considered
an argument of powiedać ‗say / speak‘, and only the contents of the complement clause are
drawn from the indicated source of information. Interpretation b., which seems the more
plausible one, assigns parenthetical status to jako powiedają ludzie ‗as people say‘, and the
complement clause explains what to ‗this‘ refers to. In this case, the parenthesis is the
27 Both scenarios are unsatisfactory with respect to the question why the perfective adverbial participle
remains respectively becomes the second base, as its anteriority is contradictory to the commentary function
of KMs.
One explanation might be language contact with German. Apart from the ADV + PTCP.PASS Cxn, in German exists
the second Cxn um es (mal) (ADV) gesagt zu haben to ha e it said o e ADV , o igi all a pu pose lause ith
the perfective verb form gesagt haben. The perfective German verb form allows to draw a parallel to the Polish
perfective adverbial participle. Although ‘a us : oti es that „ g rammatical replication im engeren
“i e fi det … e utli h ehe i i htsla is h-slavischen Kontakt statt [grammatical replication in sensu
stricto presumably takes place rather in non-Slavic – Slavic language contact – translation S.B. , the question
remains how far-fetched this explanation is, as the described replication process requires very profound
knowledge of both German and Polish and the mentioned German Cxn is less frequent than the ADV + PTCP.PASS
Cxn.
47
information source for the whole utterance, i.e. it scopes over the whole sentence. In example
(78) we are clearly dealing with a parenthesis. Again, it scopes over the whole utterance.
Thus, the syntactic disintegration that leads to the parenthetical status of the verbum dicendi
goes along with an extended scope: elements preceding the parenthesis may also be in its
scope, whereas only the (subsequent) complement clause fills the argument slot of a verbum
dicendi in a complex clause like (75).
Since the parentheses can be described as indefinite-personal clauses (ludzie ‗people‘ is just as
indefinite as the 3PL-ending of the verb without a corresponding overt subject), it is clear that
the first argument of the verbum dicendi is not identical neither with the speaker nor with the
subject of the clause into which the parenthesis is inserted.
(75) ... sposób panowania a sprawowania ludzi, jako przystoi, co samo może przywie ć
ludzie ku błogosławieństwu, a wrócić nazad on złoty wiek, o którym powiedają, że na ten
czas był, kiedy Saturnus królował.
‗ ... the way of reigning and leading people, as it befits, which is the very thing that can
lead people to beneficence, and return that golden age, about which they say, that it was at
the time when Saturnus reigned.‘
(76) ... je liż to tak ma być, jako powiedają ludzie, iż i nieprzyjaciela swego zdradzać nie
przystoi ...
a. ‗... if this really has to be so, (just) as people say that it does not befit to deceive
even one‘s antagonist …‘
b. ‗... if this really has to be so, as people say, (namely) that it does not befit to
deceive even one‘s antagonist …‘
(77) ... nakoniec, Jowisz sam, jako powiedają, królestwa swego bez niej nie mógł by dobrze
sprawować ...
‗... finally, Jupiter himself, as is said (lit. they say), could not lead his kingdom well
without her …‘
Our first example for a AP-based type of the contextualizing function comes from the 18th
century (78). Unfortunately, it already represents the final stage of development, as the
context makes clear that the speaker is the covert subject of the AP, and not the first argument
48
of the matrix verb. In other words, just as in the case of the stance-marking Cxn, the AP has
lost subject co-reference and the status of secondary predication. This allows to draw the
conclusion that with respect to these issues, the steps of syntactic development were the same
as for the speech marker and the stance-marking Cxn. Concerning valency, however, the
contextualizing Cxn differs from the stance-marking one, as it has only one argument slot.
This can be described as the result of inheriting only the semantically motivated argument
slots of the AP mówiąc ‗speaking‘, whereas the stance-marking Cxn inherits the syntactically
motivated one as well. A closer look at the semantics of the contextualizing Cxn reveals that
the respective Cxn does not indicate the source of information, i.e. of some content – as the
indefinite-personal parenthesis does – but rather the source (discourse or semantic frame) of
some lexical item which the speaker uses to convey their own information. That is, the
contextualizing Cxn is used when one content is conveyed with the help of lexical material
from two discourses – the current one produced by the speaker and another one, be it
technical as in (78) – and marks the lexical material from the ―foreign‖ discourse by scoping
over it. In contrast, the stance-marking Cxn connects to each other two variants of the same
content which both stem from the same discourse and differ only concerning the degree of
subjectivity attached to them.
(78) ... dywdyk z materii takiejż bogatej okrywał konia, cały zad aż po kostki zadnich nóg,
czyli mówiąc po rostrucharsku, aż po pętlinę.
‗ ... a caparison out of equally opulent material covered the horse, the whole croup down
to the bones of the hindlegs or, speaking in horseman‘s language (lit. in the horseman
way), down to the fetlocks.‘
(Kitowicz. Opis ...)
3.2.4. The quotative construction based on mówiąc

As has already been mentioned in section 3.2.1., the speech marking function can also be used
to mark citations. How does this function (58-59; 79) differ from the quotative Cxn (80)? The
speech or citation marker introduces new information that originates from another discourse –
in (79) a citation from the Bible is given which is then interpreted by the sermonizer. I.e.
neither the content nor its wording are conceived by the speaker of the current discourse. In
the case of the quotative Cxn, however, the speaker seeks to convey their own content by
using the wording from another discourse for this end (cf. 80-82, but also the contemporary
example 83). This allows to draw a parallel between the contextualizing and the quotative
49
Cxn, as for both the conveying of one piece of content with the help of lexical material from
two discourses is typical.
(79) [S]łyszeli cie iże rzeczonoŚ Miłuj bliznego twego a nienaźry nieprzyjaciela swego.
‗You have heard that it was saidŚ Love your neighbor and hate your enemy.‘
(RozPrzem)
The usage of another speaker‘s wording for conveying own content can already be observed
in examples from the 16th (80) and 17th centuries (81-82), where the parenthesis indicating
the source of quotation is formed with a finite verbum dicendi.
(80) Potrzeba, jako filozof rzekł, justum animatum, to jest sprawiedliwo ci żywej i
mówiącej.
‗It is required, as a philosopher said, a just mind-set, that is justness in life and speech.‘
(P. Skarga. Kazania sejmowe. Kazanie siódme.)
(81) … drugim niektórym ch<c>ąc się w tym równać, cum puderet non esse impudentem –
jako Augustyn ś. mówi.
‗... some others wishing to draw level in this respect, because it embarrasses not to be
impudent – as St. Augustinus says.‘
(Twardowski)
(82) Także Kongregacyjej W[asz]m[o ]ciów, moich M[iło ]ciwych Panów, barzo dobrze
służyć może, co ś. Bonawentura mówiŚ „Qui potentem (natalibus vel ingenio) in bono
promovet, multos iuvat. Et e converso, ipsius subversio multo<r>um est detrimentum‖ś
bo innego, który taki nie jest, „salus sibimet prodest‖, a z owych za każdego „multis, tam
propter – jako tenże przydaje – exemplum aliorum, qui inde aedificantur et provocantur
ad bonum, quam propter alia bona, quae per eum promoventur in aliis et mala
impediuntur‖.
‗Just as well, it could be of very good service to the Congregation of Well-Born, my
gentlemen, what St. Bonaventura says: Who brings forward his power (lit. powerful) (of
ships or intellect) with good intent, helps many. And on the contrary, the ruin of that
person is a harm for many‖ś because of another, who is not like this, ―the welfare is of
profit to himself only‖, and of these each one [is of profit] ―„multis, tam propter – jako
ten e przydaje – exemplum aliorum, qui inde aedificantur et provocantur ad bonum, quam
propter alia bona, quae per eum promoventur in aliis et mala impediuntur‖‘
50
(Twardowski)
As with the speech marker, Latin plays a special role. Without doubt, it is the original
language of the quotations in examples (81-82). In the case of rather longish quotations (82) it
is surely economical not to translate, if the readers – as was the case – know Latin anyway.
However, in (81) the quotation is so short that an economical motivation can be doubted28 –
not to speak of (80), where the information about the quoted philosopher is so sparse that we
cannot even infer the original language of the quotation. Therefore, it seems that in these
context Latin has a second role, namely the demarcation of narration and quotation, and thus
of the two discourses. If one considers that quotation marks and mandatory rules for their
usage developed late in the history of language (cf. Finnegan 2011: 79-95), it is not surprising
that graphical means were used to mark quotations in historical texts, especially Bible texts
(cf. Finnegan 2011: 82-84). Consequently, one could describe the resorting to Latin as means
for marking quotations as a special kind of graphical marking, as not the graphs itself, but the
arrangement of graphemes in Polish and Latin differs fundamentally. Quite notably, the AcI
gets out of use at the turn of the 18th to the 19th century (cf. Długosz-Kurczabowa & Dubisz
2006: 481), i.e. at a time when punctuation becomes normalized and direct speech as well as
verbatim quotations thus become identifiable by this means.
Unfortunately, our diachronic corpus data from the 18th century does not yield any search
results for the quotative Cxn, which makes it impossible to delineate the single steps of its
development. However, since it is formally and functionally rather close to the
contextualizing Cxn, we may assume that it developed in the same way, or even in analogy to
it.
Finally, some words are in order on the fate of the parenthesis with a finite verbum dicendi
(henceforth finite parenthesis), which we have described a predecessor of the quotative Cxn.
The data from Contemporary Polish suggests that the finite parenthesis (84-86) was not only
the predecessor, but also the competitor of the quotative Cxn, which is stepwise being
replaced by the latter. 29
(83) Ja my lę, że obaj mają swoje - mówiąc językiem Lecha Wałęsy - plusy dodatnie i
plusy ujemne.
28 furthermore, the provisional examination of one (!) edition of St. Augustinus showed that the original
wording seems to be pudet non esse impudentem, i.e. the quotation has been amended by inserting the
conjunction cum e ause a d adapti g the ood to it as to ette i teg ate it i to the host dis ou se.
29 our query jako [base="(rzec|mówić|powiedzieć)" & person=ter] gave only two relevant matches, which are
cited in examples (85-86).
51
‗I think that both have their – using Lech Wałęsa‘s wording (lit. speaking with Lech
Wałęsa‘s language) – positive pluses and negative pluses.‘
(Dziennik Polski. 2001-05-18)
(84) I rozszedł się zapach nieopisany, zapach miły Gilgameszowi, starotestamentowym
prorokom, spragnionym Orientu biznesmenom. A że – jako rzecze przysłowie - nie ma
zapachu bez dymu, objawił się i dym, wznoszący się prosto ku sufitowi.
‗An indescribable smell began to spread, a smell pleasurable to Gilgamesh, to the prophets
from the Old Testament, and to businessmen thirsty for the Orient. And since – as the
proverb says – there is no smell without smoke, smoke also appeared, ascending simply to
the ceiling.‘
(Ignacy Karpowicz. 2007. Nowy kwiat cesarza (i Pszczoły))
(85) Ja żyję w okropnym okresie, bo żądają ode mnie 20-minutowego przemówienia, czego
nigdy nie robiłam, wystąpienia w telewizji, czego do tej pory unikałam, wywiadów,
których z zasady nie udzielałam. Jednym słowem "każdemu to, na czym mu mniej zależy",
jako rzecze Rudnicki.
‗I have a terrible time, because they demand from me a 20-minute address, what I have
never done before, they demand a TV appearance, what I have avoided so far, they
demand interviews, which I have not conceded to out of principle. In one word, ―everyone
gets, what he deems least important‖, as Rudnicki says.‘
(Ryszard Matuszewski. 2004. Alfabet Ś wybór z pamięci ř0-latka.)
3.2.5. The constructions based on the infinitive

Our research hypothesis for these Cxns (henceforth infinitive-based Cxns) says that they
developed out of purpose clauses with an infinitive as predicate (henceforth infinitival
purpose clauses), whose (syntactically non-realized) first argument is coreferent with the first
argument of the superordinate clause. Therefore, it is necessary to discuss the history of this
type of purpose clause before we can go on to the development of the Cxns under
investigation. According to Bajerowa (1964: 99), the infinitival purpose clause replaced a
finite one; this process lasted about three centuries, from the second half of the 16th to the
middle of the 19th century. Information about its peak variesŚ Jodłowski (cited in Bajerowa
1964: 100) identifies the 17th century as the period of the most intensive change processes,
whereas Bajerowa‘s empirical data show that the development is at its peak after 1770
(Bajerowa 1964: 100). The latter findings coincide with those of Długosz-Kurczabowa &
Dubisz (2006: 480), which leads us to take them as point of departure for the interpretation of
52
our data. One possible explanation for the development of infinitival purpose clauses is
German influence (Bajerowa 1964: 99).
Given the history of infinitival purpose clauses as such, it is not too surprising that our corpus
data from the 16th century yields no examples of the infinitive-based Cxns. Instead, we find
two other Cxns that fulfill its function. They use the same verba dicendi as the infinitive-
based Cxns in three inflectional forms – the 1SG.COND (86), the 1SG.PRS (87-89) and an
ambiguous form (87; 90-91) that can be analyzed as the old form of the adverbial participle in
the NOM.SG.M on –ę or as 1SG.PRS. In our data, the ambiguous form occurs both with (90—
91) and without (87) the complementizer iż. The former makes the interpretation as 1SG.PRS
more likely, the latter the interpretation as AP. However, we also cannot exclude that we are
in fact dealing with an AP – the more so, as non-coreferential APs in the speech marking
function, often modified by tak(o) ‗so‘, are attested – that was misinterpreted as 1SG.PRS and
therefore furnished with the complementizer iż. The fact that Czech features the AP-based
Cxn tak íkajíc ‗so to say‘ supports the assumption that Polish rzekę has to be considered an
AP. Due to its formal conflation with the 1SG.PRS, the old AP form on -ę gets out of use (cf.
Długosz-Kurczabowa & Dubisz 2006: 322); Weiss (1984: 175) notes the middle of the 16th
century, i.e. the period our examples stem from, as the date of their extinction. Therefore, it is
possible that the Cxn based on the 1SG.PRS came into being as analogy to the alleged 1SG.PRS
rzekę, which got out of use due to its ambiguity. The conditional form in (86), in turn, forms
part of a purpose clause und thus may be considered a predecessor of the infinitival purpose
clause.
Note that the AP-based Cxns (87; 90-91) already occur in the positions where modern hedges
and fillers occur, namely directly before (87; 91) or after (90) the element they scope over.
(86) ... to Jest rzecz tak trudna, iż ledwo bych nie rzekł nie można ku uczynieniu.
‗... this thing is so difficult, that I would rather say (lit. that I would hardly not say)
impossible to accomplish.‘
(87) ... ono nie rzekę zuchwalstwo, ale błazeństwo było, kiedy Alexander wielki ... zapłakał.
‗... it was not to say a danger, but stupidness, when Alexander the żreat … started to cry.‘
53
(88) ... na to powiesz, iż to w nich Duch więty sprawowałś prawda jest, ale ja zaś tak
powiedam30, iż to najwiętsza i najchwalebniejsza cnota, o której pan Bóg duchem swym
wiadectwo daje, bo we złe serce duch Boży nie wchodzi.
‗... you will say on this, that the Holy żhost has acted in them; it is true, but I say so, that
it is the amplest and most laudable virtue, about which God the Lord gives testimony
through his ghost, because the Holy żhost does not enter evil hearts.‘
(89) Do tego jeszcze przyłożę koszt a utraty ... bo chocia sie tak zda, iżby nic na tym nie
należało, ale ja tak powiedam, iż takowe zbytki niszczą królestwa;
‗To this I still add expenditures and losses … because although it seems, that nothing
depends on this, but I say so, that such luxuries ruin a kingdomś‘
(90) ... to takie wymy lone (iż tak rzekę) dworzaństwo ...
‗... such a devised (so to say (lit. that so saying)) caste of courtiers …‘
(91) ... radbych ja wiedział, je li to (iż tak rzekę) ochmistrzowanie dworzaninowe ma sie
stąd począć ...
‗... I would be glad to see, if these (so to say (lit. that so saying)) patronized courtiers
had to start from that …‘
Our corpus size does not allow to argue on quantitative grounds, but it is nonetheless notable
that the ambiguous form rzekę is attested only in the Cxn preceding the infinitive-based Cxn.
Therefore it is possible that the following scenario has taken place: Until the 16th century a
Cxn based on the old AP on -ę has developed, which fulfills the functions of the modern
infinitive-based Cxn. The AP is non-coreferential and modified by tak(o) ‗so‘ or nie ‗not‘.
The missing co-referentiality as well as the modification by tak(o) ‗so‘ are also characteristic
of the speech marking Cxn, but the two Cxns are disambiguated by the complementary
distribution of the AP suffixes –ę and –ąc. Additionally, there exist two more Cxns that may
be considered the predecessors of the modern infinitive-based Cxn. The first one, based on the
1SG.PRS, is analogous to the second reading of the form on –ę; the second one is in the
conditional and thus a predecessor of the infinitival purpose clause (the intersections are
30 apa t f o sa powieda is also attested ith the ea i g thi k, o side . Ho e e , this does ot
weaken our argumentation: many hedges and fillers based on verba cogitandi occur in various languages, e.g.
E glish I thi k , so the i te p etatio of powieda as verbum cogitandi would not contradict its status as point
of departure for the development of a y / y / że y tak powiedzie / rze with the hedging and filler function.
54
illustrated in Chart 1). Both are theoretically able to function as parenthesis. When the old AP
on –ę gets out of use because of its ambiguity, the stage for a new Cxn is prepared – but what
is the decisive factor that makes the conditional and thus finally the infinitive-based Cxn
succeed?
COND predecessor of infinitival purpose

AP form -ę PRS 1SG.
clause
Chart 1. Intersections between attested preceeding forms of the infinitive-based Cxn.
Unfortunately, our diachronic corpus data cannot give answer to this question, as it offers no
relevant matches. Therefore, we have to track the hints Bajerowa (1964) gives. As mentioned
above, the infinitival purpose clause evolves between the middle of the 16th and the middle of
the 19th century, with a peak of most intensive development in the period after 1770. German
influence is considered one possible motivation factor for the development of the infinitival
purpose clause. Thus the question arises whether it is indeed plausible that language contact
played a role, and whether this assumption also applies to the development of the infinitive-
based Cxn. In answering this question, we will take as orientation the scheme for detecting
calques proposed in Weiss (2009).
For the 18th and 19th century, apart from German the other important contact languages were
French and Russian. Polish had only literacy contact with French, whereas the contact with
German and Russian was face-to-face. Furthermore, it is highly probably that German often
functioned as mediator language for borrowings from French (cf. Rabus 2013: 218); Raecke
(1994: 226) even states that no imitation of French syntax has taken place during the 18th
century. These facts make German influence on the development of the infinitival purpose
clause more likely than French influence. If we take into consideration the structural
characteristics of the German (92-94), French (95-96) and Russian equivalents (97-98) of
Polish aby / by / żeby tak powiedzieć / rzec ‗so to say / in order to say so‘ and aby / by / żeby
nie powiedzieć / rzec ‗in order not to say‘, it turns out that the żerman equivalents are
structurally closest to the Polish Cxns, followed by Russian and French: German um … zu ‗in
order to‘ is the direct equivalent of Polish aby / by / żeby ‗in order to‘, whereas Żrench pour
‗for‘ equals the Polish preposition dla ‗for‘, which is to be followed by a verbal noun instead
of the infinitive.
(92) so-zu-sagen
55
so-to-say.INF
‚so to sayÄ
(93) um es so zu sagen
in_order it-NOM so to say-INF
‗so to say (lit. in order to say it so)‘
(94) um nicht zu sagen
in_order not to say-INF
‗not to say (lit. in order not to say)‘
(95) pour ainsi dire
for so say-INF
‗so to say‘
(96) pour ne pas dire
for not NEG say-INF
‗not to say‘
(97) tak skazat‘
so say-INF
‗so to say‘
(98) čtoby ne skazat‘
in_order-COMPL not say-INF
Although the same structural differences exist between Russian and French, Russian is known
to have taken French as model (cf. Birzer 2012). That the same is also true for Polish can be
excluded for the differing degree of French influence on Polish and Russian. Rabus (2013:
219) describes this difference as follows:
Der strukturelle Einfluss des Französischen auf das Polnische war im Hinblick auf seine Nachhaltigkeit
insgesamt geringer als der auf das Russische, wo dieser Einfluss die russische Syntax europäisierte. Beim
französisch-polnischen Kontakt unterstützten häufig die nachhaltigen Erscheinungen lateinische
Phänomene katalytisch.
[With respect to endurability, the structural influence of French on Polish was all in all lower than on
Russian, where this influence Europeanized Russian syntax. In the French-Polish contact the endurable
phenomena supported Latin phenomena as catalyzers. Translation – S.B.]
The Cxns under investigation confirm these findings: firstly, they are not influenced by Latin,
and secondly, their Russian equivalents are representatives of the pattern ADV + skazat‘-INF
56
that was semi-productive in the second half of the 18th and the first half of the 19th century
and included many semantic equivalents of the corresponding French pattern (cf. Birzer 2012:
chapter 4.2). Polish, in contrast, displays only the two Cxns aby / by / żeby tak powiedzieć /
rzec ‗so to say / in order to say so‘ and aby / by / żeby nie powiedzieć / rzec ‗in order not to
say‘ which are structurally identical to their żerman equivalents.
Therefore, we can assume that German influence played the decisive role in the development
of the infinitive-based Cxns.
4. Conclusion
The aim of this chapter is to assess the role language contact played in the development of the
following four functional types of KM based on infinite forms of the verba dicendi mówić
‗speak‘, powiedzieć ‗say‘ and rzec ‗say‘: the stance-marking function with the prototypical
Cxn ADV + mówiąc, the contextualizing function with the prototypical Cxn mówiąc +
ADJ.INSTR + NOUN.INSTR, the quotative function with the prototypical Cxn mówiąc +
NOUN.INSTR + PROPER_NOUN.GEN, as well as the infinitive-based Cxns aby / by / żeby tak
powiedzieć / rzec ‗so to say / in order to say so‘ and aby / by / żeby nie powiedzieć / rzec ‗(in
order) not to say‘ with hedging and filler as primary functions. We will carry out this
assessment at first from an inner-Polish perspective and then, in a second step, from an outer
perspective that will incorporate the analysis of similar Cxns in Russian (Birzer 2012) and
Croatian (Birzer accepted b).
For all discussed Polish Cxns we have to assume the adverbial participle as point of departure,
which gradually loses subject co-reference with the first argument of the matrix sentence and
thus the status of a secondary predication.
Until the 16th century we observe the development of a speech marker that is applied to mark
both direct and reported speech and, in a later step, also citations. Already at rather early
stages, structural parallels between the speech marking in Latin original texts and their Polish
translations are obvious especially in clerical texts. Until the 16th century a Cxn based on the
old AP on -ę has developed, which fulfills the functions of the modern infinitive-based Cxns.
The AP on -ę is non-coreferential and modified by tak(o) ‗so‘ or nie ‗not‘. The missing
coreferentiality and modification by tak(o) ‗so‘ are also characteristic of the speech marking
Cxn, yet the two Cxns are disambiguated by the complementary distribution of the AP
suffixes –ę and –ąc. This may be considered a morphological means of disambiguation.
57
On the other hand, at the older stages of Polish, just as in other languages (cf. Večerka 2002Ś
416-423 for Old Church Slavonic or Daiber 2009 for Russian Church Slavonic), the border
between direct and reported speech was rather unclear. Latin, in contrast, has the AcI as
means to encode reported speech; the infinitive is clearly distinct from the finite verb forms to
be expected in direct speech. In the period of macaronism, Polish replicates the syntactic
pattern of the Latin AcI - sometimes using Latin, sometimes Polish lexical material – and thus
acquires a means to differentiate between direct and reported speech. Quite interestingly, in
the quotative Cxn, which develops out of the speech marking Cxn, Latin insertions are also
recurred to. This is not too surprising if the original language of the quotations is Latin, but
this is incurred also in contexts where the information about the quoted author is so sparse
that the original language of the quotation cannot even be inferred. Again, Latin has a
distinctive function; in the given case it demarcates the narration (in Polish) and the quotation
(in Latin). Quite notably, the AcI gets out of use at the turn of the 18th to the 19th century, i.e.
at a time when punctuation becomes normalized and direct speech as well as verbatim
quotations thus become identifiable by this means.
The literary ouput of the 16th century is usually labeled with the keyword macaronism, yet
Dubisz distinguishes between ―Latinzim to element językowy (najczę ciej wyraz)
zapo yczony z łaciny i przystosowany do polskiego system językowegoś makaronizm to obcy
element językowy (w naszym przypadku – łacinski), występujący w polskim tek cie w
oryginalnej obcojęzycznej postaci językowej [Latinism is a language item (most often a word)
that is borrowed from Latin and integrated to the Polish language system; makaronism is a
foreign language lement (in our case – a Latin one) that occurs in the Polish text in its original
foreign linguistic form – translation S.B.]‖ (2007Ś 10). The AcI as pattern replication can
neither be labeled a Latinism nor a macaronism in Dubisz‘s sense, whereas the usage of Latin
without morphological integration into Polish corresponds to a macaronism. In the
terminology of contact linguistics, the latter phenomenon can be described as a functional
code-switch (cf. Riehl 2004: 23) that serves the demarcation of two text segments authored by
different persons.
Latinisms, on the other hand, can be considered a case of matter replication. The expansion of
the prototypical stance marking Cxn ADV + AP is the result of the intricate interplay between
catalyzation and matter replication (in Sakel‘s (2007) sense). Most probably, the structural
and functional parallels between a semantically corresponding Latin parenthetical Cxn and the
Polish ADV + AP Cxn lead to matter replication from Latin, which resulted in rising type
frequency. Therefore, the Latin parenthetical Cxn as such was the catalyzer for the expansion
58
of the Polish ADV + AP Cxn, and the concrete types of the Latin Cxn served as resource for
mattern replication.
Finally, concerning the modern infinitive-based Cxns, pattern replication from German has to
be assumed for the development of the infinitival purpose clause, which serves as basis of the
respective Cxns. Additionally, the two infinitive-based Cxns aby / by / żeby tak powiedzieć /
rzec ‗so to say / in order to say so‘ and aby / by / żeby nie powiedzieć / rzec ‗(in order) not to
say‘ themselves display a direct syntactic and semantic parallel to their German counterparts
sozusagen ‗so to say‘ and um nicht zu sagen ‗(in order) not to say‘. Therefore, one may
assume that the pattern replication of the infinitival purpose clause served both as syntactic
model and catalyzer for the semantically motivated replication of the two infinitive-based
Cxns.
To summarize the inner-Polish perspective on the role of language contact, the development
of the Cxns under investigation involved both pattern and matter replication. The pattern
replication processes can be differentiated into structural and semantic ones. The result of the
syntactic pattern replications can be roughly described as the crystallization of functions,
whereas the semantic pattern replication and the matter replications always went along with a
catalyzing effect.
Let us now turn to the outer perspective. Just like Polish, Russian and Croatian have several
metatextual Cxns based on verba dicendi, among them several ones (in each language) based
on the AP. Due to this distribution, the AP-based Cxns can be identified as genuinly Slavonic.
One of them is the stance-marking ADV + AP Cxn, which is semi-productive and unchallenged
by competing replicated Cxns in Polish. In contrast, language contact lead to competing
stance-marking Cxns in both Russian and Croatian. In the case of Russian, language contact
with French resulted in the pattern replication of the ADV + INF Cxn already described in
section 3.2.5. Since the ADV + AP Cxn already existed in Russian at that time, there was no
―functional gap‖ in Russian (which in the older literature on borrowing has often been
assumed to be the motivation for replications). In Modern Russian the replicated pattern is no
more productive, but the genuinly Slavonic ADV + AP Cxn is semi-productive, as in Polish.
Thus the replicated pattern can be considered as having catalyzed the expansion of the already
existent ADV + AP Cxn.
The Croatian setting is different in the sense that it is the only of the three investigated
languages that has preserved the ―bare‖ AP as speech marker, which to a certain degree
results in the formal underspecification between the speech marking and the stance-marking
function. Language contact with German resulted in the syntactic replication of the stance-
59
marking Cxn ADV + PTCP.PASS (e.g., Croatian iskreno rečeno-PTCP.PASS in analogy to German
ehrlich gesagt-PTCP.PASS ‗frankly speaking (lit. spoken)‘). A functional motivation for the
replication process can be excluded, as the originally Slavonic Cxn ADV + AP already existed.
However, the availability of two distinct syntactic patterns for marking the same function
prepared the ground for dissolving the aforementioned formal underspecification between the
speech marking and the stance-marking function. Consequently, in Modern Croatian the AP is
preferably used as speech marker, i.e. has retained (or rather returned to) its initial function,
whereas the replicated ADV + PTCP.PASS Cxn has become the semi-productive means for
forming stance-marking expressions. We may thus speak of a crystallization of functions, as
every function has become linked to a specific Cxn.
If we compare the findings from Polish, Russian and Croatian, we may conclude the
following: in language contact situations, replications take place independent of ―gaps‖ in the
language system, but their further fate seems to be linked to the structural specificities of the
given system. If the target language already features a functionally identical Cxn, the
replicated Cxn seems to be confined to the role of a catalyzer for the expansion of the ―native‖
Cxn, as the stance-marking Cxns in Polish and Russian show. However, if the Cxn is
replicated in a situation of functional underspecification, it leads to a crystallization of
functions, as each of the functions becomes linked to one specific Cxn. The functionally
specific replicated Cxn in turn may get out of use again if another marker for the same
function evolves (as seems to have been the case with the Polish AcI Cxn as marker of
reported speech) or become firmly settled as the only (or at least preferable) marker of the
given function, as was the case with the Croatian ADV + PTCP.PASS Cxn for stance-marking.
Żurthermore, our data support Rabus‘s (2013: 55) finding that ―[g]rammatical replication im
engeren Sinne findet somit vermutlich eher im nichtslavisch-slavischen Kontakt statt
[grammatical replication in sensu stricto presumably takes place rather in non-Slavonic-
Slavonic language contact – translation S.B.]‖, as all discussed replicated structures stem from
non-Slavonic languages.
References
Bajenowa, Irena. 1964. Kstałtowanie się systemu polskiego języka literackiego w XVIII
wieku. Wrocław et al.: Ossolineum.
60
Birzer, Sandra. (accepted a) Formal unity and functional diversity: a corpus-linguistic
approach to Russian and Polish adverbial participles with the meaning 'speaking'
between discourse and grammar.// Żried, Mirjam / Lehečková, źva (eds.)
Conjunctions vs. Contextualizers. John Benjamins. (working title) (Ms. 40 pages)
Birzer, Sandra. (accepted b) Historical development and contemporary usage of discourse
structuring elements based on verba dicendi in Croatian // żrković-Major, J. & B.
Hansen (eds.) Diachronic Slavonic Syntax. Mouton de Gruyter. (Ms. 53 pages)
Birzer, Sandra. 2012a. From subject to subjectivity: Russian discourse structuring elements
based on the adverbial participle govorja ‗speaking‘. // Russian Linguistics, 36/3. 221-
249.
Birzer, Sandra. 2012b. Von der (in)direkten Rede zum Kommentar. Verba dicendi in
griechischen und altkirchenslavischen Konstruktionen der Redewiedergabe. // Hansen,
B. (Hrsg.) Diachrone Aspekte slavischer Sprachen. (Slavolinguistica 16). München /
Berlin: Otto Sagner. 197-209.
Birzer, Sandra. 2013. Constructions based on the Russian adverbial participle govorja and
their function in discourse and grammar. // Kemgpen, S. et al. (eds.) Deutsche
Beiträge zum 15. Internationalen Slavistenkongress Minsk 2013. 67-77.
Birzer, Sandra. 2015. Generally speaking, connectivity and conversation management
combined. The functions of Russian voobšče govorja and Polish ogólnie mówiąc
‗generally speaking‘. // Russian Linguistics, 39/1. DOI 10.1007/s11185-014-9143-0.
Czapiga, Zofia. 2006. Rosyjskie operatory metatekstowe i ich polskie odpowiedniki. Rzeszów:
Wydawnictwo Uniwersytetu Rzeszowskiego.
Daiber, Thomas. 2009. Direkte Rede im Russisch-Kirchenslavischen. Zum pragmatischen
Wert des jako recitativum. in: Besters-Dilger, J., Rabus, A. (Hrsg.): Text – Sprache –
Grammatik. Slavisches Schrifttum der Vormoderne; Festschrift fur Eckhard Weiher.
München, 363-386.
Dehé, N. / Kavalova, Yo. (eds.) 2007. Parentheticals. Amsterdam: John Benjamins.
Długosz-Kurczanowa, Krystynaś Dubisz, Stanisław. 2006. Gramatyka historyczna języka
polskiego. Warszawa: WUW.
Dubisz, Stanisław. 2007. Wpływy łaciny na język polski. // Poradnik językowy, 5/2007, 3-13.
Feret, Andrzej S. 2005. Das Partizip im Deutschen und Polnischen. Frankfurt /M. – Berlin –
Bern – Bruxelles – New York – Oxford – Wien.
Finnegan, Ruth. 2011. Why do we quote? The culture and history of quotation. Cambridge:
Open Book Publishers. (http://www.openbookpublishers.com/reader/75)
61
Fraser, Bruce. 2006. Towards a theory of discourse markers. in: Fischer, K. (ed.) Approaches
to Discourse Particles. Amsterdam et al. 189-204.
Gries, S. T.; David, C. V. 2007. This is kind of / sort of interesting: variation in hedging in
English. In: Pahta, P. et al. (Eds.), Studies in variation, contacts and change in
English. Volume 2: Towards multimedia in corpus studies.
http://www.helsinki.fi/varieng/journal/volumes/02/gries_david/
Grochowski, Maciej et al. 2014. Słownik gniazdowy partykuł polskich. Kraków: Polska
Akademia Umiejętno ci.
żrochowski, Maciej. 1983. Metatekstowa interpretacja parentezy. inŚ Dobrzyńska, T. / Janus,
E. (ed.) Tekst i zdanie. Tom studiów. Wrocław. 247–258
Grochowski, Maciej. 1986a. O metapredykatywnej funkcji niektórych wyra eń partykułowo-
przysłówkowych w strukturze tekstu. inŚ Dobrzyńska, T. (ed.) Teoria tekstu. Zbiór
studiów. Wrocław. 140-
Grochowski, Maciej. 1986b. Polskie partykuły. Składnia, semantyka, leksykografia. Wrocław.
Grochowski, Maciej. 1998. Funktionswörter im Polnischen. Oldenburg.
Güldemann, Tom. 2012. Thetic speaker-instantiating quotative indexes as a cross-linguistic
type. In I. Buchstaller & I. van Alphen (Eds.), Quotatives: cross-linguistic and cross-
disciplinary perspectives (117-142). Amsterdam: John Benjamins.
Hansen, Björn. 1996. Zur Grammatik von Referenz und Episodizität. (Specimina philologiae
slavicae Supplementband 46) München: Verlag Otto Sagner.
Hansen, Björn. 2010. Mood in Polish. In: Rothstein, Björn; Thieroff, Rolf (eds.) Mood in the
Languages of Europe. Amsterdam: John Benjamins. 342-357.
Kaltenböck, G. / Heine, B. / Kuteva, T. 2011. On thetical grammar. Studies in Language,
35/4, 852–897.
Kara , Halina. 1996. Rusycyzmy słownikowe w polszczyźnie okresu zaborow. Warszawa:
Elipsa.
Karłowicz, Jan. 1984. Podręcznik czystej polszczyzny dla Litwinow i PetersburszczanŚ do
druku przygotowała źl bieta Smułkowa. In Janusz Rieger & Wiaczesław Werenicz
(eds.), Studia nad polszczyzną kresową: Tom III (Prace Językoznawcze 107.), 33-81.
Wrocław [etc.].
Kaupu , A. & ź. Smułkowa. 1984. Nieznane prace Jana Karłowicza o języku polskim na
Wileńszczy nie. In Janusz Rieger & Wiaczesław Werenicz (eds.), Studia nad
polszczyzną kresową: Tom III (Prace Językoznawcze 107.), 82-90. Wrocław [etc.].
62
Keipert, H. 1988. Sprachprobleme der makkaronischen Dichtung in Polen. // Die Welt der
Slaven, 33, 354-388.
Kędelska, źl bieta, Kwilecka, Irena & Arleta Łuczak (eds.) 1999-2009. Słownik Bartłomieja z
Bydgoszczy: wersja polsko-łacińska. WarszwaŚ Slawistyczny O rodek Wydawn.
Klemensiewicz, Zenon. 71999. Historia języka polskiego. Warszawa: Wydawnictwo naukowe
PWN.
Marchołt http://daten.digitale-
sammlungen.de/~db/0002/bsb00022012/images/index.html?id=00022012&fip=xsyzts
xdsydenxsxdsydeneayasdasyzts&no=8&seite=6
Matras, Yaron 1998. "Utterance modifiers and universals of grammatical
borrowing." Linguistics 36/2: 281 - 331.
Mączyński, Jan (ed.). 1973. Lexicon Latino-Polonicum ex optimis Latinae linguae
scriptoribus concinnatum. Köln: Böhlau.
Mendoza, Imke. 2009. Anaphorische Mittel: Konnexion. In S. Kempgen et al. (Eds.), Die
slavischen Sprachen. Ein internationales Handbuch zu ihrer Struktur, ihrer
Geschichte und ihrer Erforschung. Band 1 (Handbücher zur Sprach- und
Kommunikationswissenschaft, 32.1). Berlin, New York. 982–990.
Meyer, Roland. 2011. The History of Null Subjects in North Slavonic. A Corpus-Based
Diachronic Investigation. Regensburg. (Ms. Habilitation Thesis, 274 pages).
Moszyński, L. 1994. O staropolskim i staroczeskim przekładzie pierwszych zdań Księgi
Rodzaju z oryginału hebrajskiego (Biblia Budnego i Kralicka). // Kucała, M. /
Krą yńska, Zd. (red.) Studia historycznojęzykowe I. Kraków, 149-156.
Mymerus, Franciscus. 1997 [re-edition]. Francisci Mymeri Dictionarium trium linguarum.
Kraków, Collegium Columbinum.
Nowowiejski, Bogusław. 2007. Wpływy języka niemieckiego na polszczyznę w zakresie
słownictwa. // Poradnik językowy, 5/2007, 14-24.
O óg, K. 1991. Elementy metatekstowe ze składnikiem mówię w polszczyźnie mówionej.
Język a kultura, 183-194.
O óg, K. 1990. Leksykon metatekstowy współczesnej polszczyzny mówionej. Wybrane
zagadnienia. Kraków.
Pisarkowa, Krystyna. 1984. Historia składni języka polskiego. Wrocław et al.Ś Ossolineum.
Porayski-Pomsta, Józef. 2007. Zapo yczenia leskykalne z języka francuskiego we
współczesnej polszczy nie. // Poradnik językowy, 5/2007, 54-65.
63
Rabus, Achim. 2013. Die Rolle des Sprachkontakts fur die slavischen (Standard-)Sprachen
(unter besonderer Berucksichtigung des innerslavischen Kontakts). Freiburg / Br. (Ms.
Habilitation thesis, 467 pages).
Raecke, Jochen. 1984. Das 18. Jahrhundert im polnischen Satzbau. In: Mehlig, Hans Robert
(ed.) Slavistische Linguistik 1993. Referate des XIX. Konstanzer Slavistischen
Arbeitstreffens. Kiel 21.-23.9.1993. München: Otto Sagner. 225-249.
Rathmayr, Renate. 1985. Die russischen Partikeln als Pragmalexeme. München.
Riehl, Claudia. 2004. Sprachkontaktforschung. Eine Einführung. Tübingen: Niemeyer.
Sakel, Jeanette. 2007. Types of loan: Matter and pattern. in: Matras, Y., J. Sakel (eds.):
Grammatical Borrowing in Cross-Linguistic Perspective. Berlin / New York, 15-29.
Schiffrin, Deborah. 1987. Discourse markers. Cambridge.
Sękowska, źl bieta. 2007. Wpływ języka angielskiego na słownictwo polszczyzny ogólnej. //
Poradnik językowy, 5/2007, 44-53.
Słownik Polszczyzny XVI wieku. 1966- . Red. Bąk, St. et al. Wrocław – Warszawa – Kraków.
Szlifersztejnowa, S. 1968. Bierne czasowniki zaimkowe (reflexiva) w języku polskim.
Wrocław et al.Ś Ossolineum.
Traugott, E. C.; Dasher, R. B. (2002). Regularity in semantic change. Cambridge.
Večerka, Rudolf. 2002. Altkirchenslavische (altbulgarische) Syntax. Band IV. Die
Satztypen:der zusammengesetzte Satz. Freiburg i. Br.: Weiher.
Wajszczuk, J. O metatek cie. Warszawa: Katedra Lingwistyki Formalnej UW.
Weiss, Daniel. (in preparation). Dеputаty ljubjat citаty: ssylki nа ksеnоtеksty v Gоsdumе.
Weiss, Daniel. 1977. Syntax und Semantik polnischer Partizipialkonstruktionen im Rahmen
einer generativ-transformationellen Sprachbeschreibung. Bern: Lang.
Weiss, Daniel. 1982. Deutsch-polnische Lehnbeziehungen im Bereich der Passivbildung. In:
Reißner, E. (Hrsg.): Literatur- und Sprachentwicklung in Osteuropa im 20.
Jahrhundert. Ausgewählte Beiträge zum Zweiten Weltkongreß für Sowjet- und
Osteuropastudien, Berlin, 197-218.
Weiss, Daniel. 1984. Kongruenz vs. Kongruenzlosigkeit: Zur typologischen Entwicklung des
Polnischen. In: Zeitschrift für Slavische Philologie 44, 144-192
Weiss, Daniel. 2009. Mögliche Argumentationen zum Nachweis von Calques am Beispiel der
polnischen Modalverben. In: Scholze, Lenka; Wiemer, Björn (Hg.), Von Zuständen,
Dynamik und Veränderung bei Pygmäen und Giganten. Bochum: Universitätsverlag
Dr. N. Brockmeyer. 129-153.
64
Wierzbicka , Anna. 1971. Metatekst w tek cie, wŚ M. R. Mayenowa (red.), O spójno ci
tekstu, WrocławŚ Ossolineum. 105–121.
Wołk, Mariola. 2007. Co mówimy, mówiąc powiedzmy? // Poradnik językowy, 1/2007, 5-19.
abowska, Magdalena. 2009. Hierarchia wyra eń metatekstowych. // Linguistica
Copernicana, 2/2009, 179-189.
65

Assessing The Role of Pattern and Matter

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Assessing The Role of Pattern and Matter

Uploaded by

Copyright:

Available Formats

Assessing the role of pattern and matter replication in the development of Polish

komentarze metatekstowe based on infinite verba dicendi

Polish employs a whole range of deverbal komentarze metatekstowe (metatextual comments,

1. Defining the research object

(7) Szanse są więc niewielkie, by nie powiedzieć zerowe.

(4i) Janek nie odpowiada mojej koncepcji dobrego przyjaciela.

[d]ie Bedeutung bzw. Funktion von multifunktionalen koordinierenden Konjunktionen

(11) Øsp Szczerze mówiąc, Adam jest głupi.

(14) Zadzwoniłamj, żeby ci powiedziećj,

2. Functions in Contemporary Polish

(31) Chmielowski interpretuje go [aktu II dramatu Wyspiańskiego – S.B.] — by tak

3.1. Used data

3.2. Historical development

3.2.1. Marking of direct and indirect speech

The first development to be mentioned is the insertion of cataphoric expressions before or

(51) POL A tak prawdziwie jest od niektórego

(52) POL I dlatego, gdy Kallistena filozofa przeto

Immo ego, inquit, [miserum et

(54) Gdy widziała Alexandra [być w tak wielkim

(69) .... wszystka kompanija – verius-ADV dicam-CONJ.PRS.1SG – pouciekała ...

(73) A to powiedziawszy wstał i onej rozmowie uczynił koniec.

Table 3. Token frequencies for powiedziawszy-based types in comparison to mówiąc-

26 the used corpus query was [orth=",|.|-"] [] [orth=powiedziawszy] [orth=",|.|-" orth!=że

3.2.3. Development of the contextualizing Cxn based on mówiąc ‘speaking’

3.2.4. The quotative construction based on mówiąc

3.2.5. The constructions based on the infinitive

COND predecessor of infinitival purpose

Chart 1. Intersections between attested preceeding forms of the infinitive-based Cxn.

You might also like