The Polish complementiser jakoby : a relevance-theoretic account

This paper analyses the Polish complementiser jakoby within the framework of relevance theory. It suggests that those environments in which jakoby is licensed, namely certain indirect-speech-type constructions and clauses embedded under inherently negative predicates, such as zaprzeczyć (“to deny”) and nieprawdą jest (“it is untrue”), have in common that they can be conceived of in metarepresentational terms. Furthermore, it argues that jakoby encodes procedural meaning which restricts it principally to these types of environment. Unlike że (“that”), which can always be substituted for it, jakoby constrains the range of attitudes towards the embedded proposition that can be implied contextually, blocking interpretations on which this proposition is understood to be endorsed, while encouraging the recovery of evaluative stances such as scepticism, doubt and rejection. We show that jakoby can be selected from the point of view of the reporting voice, in which case it receives a global interpretation, or that of the matrix subject; this yields a local interpretation.


Introduction
The Polish complementiser that corresponds most closely to English that, French and Spanish que and German dass in terms of its syntactic function is że.In some cases, substituting jakoby for że still results in a well-formed utterance: 1 1 In each case the original has jakoby; że was added by me.
3 Superscript "pf." indicates that the aspect of the verb is perfective.Superscript "impf."will be used in the case of imperfective verbs.Translations of Polish words will be given only on their first occurrence in each section.If __ 165 __ Jakoby also occurs as a hearsay particle, which corresponds approximately to English "supposedly", as in ( 6): (6) Do Polski przyjechał w 1937 r., w czasie okupacji był jakoby partyzantem, a w 1945 żołnierzem batalionu morskiego.
("He came to Poland in 1937, during the occupation he was supposedly a partisan, and in 1945 he was a soldier in the naval battalion.")("Polityka", no. 2359("Polityka", no. , 20.07.2002) Hearsay jakoby exists in a paradigmatic relationship with three other particles, namely podobno, ponoć and rzekomo, thus a full consideration of jakoby as a particle would require an investigation of the complexities of this relationship (discussed by Wiemer, 2006;Stępień 2008;2010a;2010b;Żabowska, 2008;Socka, 2010;Grochowski, Kisiel, Żabowska, 2014 andWiemer, Socka, 2017a;2017b, among others).Although some references to hearsay jakoby will be unavoidable, a comprehensive treatment of it lies outside the scope of this paper.
("It is already known that it will start at 17:00 in the conference room of Goleniów District and Town Hall.") ("Gazeta Goleniowska", 3.11.2006)Furthermore, why would a speaker choose to employ jakoby instead of że in contexts that permit this opposition to be played out?The consensus in the literature is that jakoby casts some doubt on the veracity of the embedded proposition.For Wiemer, hearsay jakoby has both an evidential/reportative component and an epistemic one.The evidential component ensures that the proposition in its scope is understood to be attributed to text-external voices and is glossed as "I want to say what someone else says" (Wiemer, 2006: 43), while the epistemic one is glossed as "I don't say I know that P" and "I think that P can be not true" (Wiemer, 2006: 43).This latter element is the same when jakoby functions as a complementiser.How it is projected in such cases depends on the nature of the matrix predicate.Some predicates, e.g. the negative zaprzeczać/zaprzeczyć or the nominal plotka ("piece of gossip"), "signal that the clause introduced by jakoby is judged as false or doubtful" (Wiemer, 2006: 44).Here, Wiemer presumably also has other inherently negative predicates in mind, such as nieprawdą jest.However, in the case of epistemically neutral predicates (he cites słyszeć impf.("to hear") and wersja ("version")) "this evaluation remains, as it were, open and can be triggered by other factors, e.g. by a negatively qualifying predicate" (Wiemer, 2006: 44).Stępień (2010a) also draws parallels between the epistemic contribution of hearsay and complementiser jakoby, applying some somewhat questionable reasoning to suggest that a verb is mentioned during discussion but without reference to a specific preceding or following example, then both aspect forms will be given: imperfective/perfective, e.g.zaprzeczać/zaprzeczyć.__ 166 __ complementiser jakoby is simply a contraction of że jakoby, i.e. hearsay jakoby (discussed by Wiemer, 2010).Like Wiemer, she proposes that hearsay jakoby, along with podobno and rzekomo, has both an evidential and an epistemic component; the latter concerns the reporting voice's doubts regarding the truth of the reported proposition.Through a series of tests, mostly involving embedding, Stępień suggests that while the evidential and epistemic functions of podobno have equal status, the evidential components of jakoby and rzekomo are subordinate to their epistemic ones.Evidence for this is that the evidential component is not projected when jakoby occurs in a subordinate clause which is both embedded under a non-factive verb of saying and introduced by że.
For Jędrzejowski, jakoby as a complementiser "clearly does not require any degree of speaker commitment" (Jędrzejowski, 2020: 104), and "contributes a dubitative component" since it can be used by the speaker "to distance herself/himself from the content of the reported proposition" (Jędrzejowski, 2020: 105).Jędrzejowski also considers the licensing conditions of jakoby and observes that those verbs that most frequently take a jakoby-clause are Karttunen's verbs of one-way communication (Karttunen, 1977), such as twierdzić/stwierdzić ("to claim") and sugerować/zasugerować ("to suggest").However, Jędrzejowski's only reference to inherently negative predicates is his inclusion of zaprzeczać/zaprzeczyć and dementować/zdementować (also "to deny") among this class of verbs, thus the association of jakoby with other negative predicates remains unaccounted for.
The aim of this paper is to suggest how the semantics and pragmatics of the complementiser jakoby may be analysed within the framework of relevance theory.We shall suggest that examples (1) to (4) above (as well as (5)) have in common that they are metarepresentational and shall suggest that jakoby can be analysed as a marker of a type of metarepresentational language termed interpretive use by relevance theorists.Linguistic expressions associated with such use have been identified in languages as diverse as Japanese (Itani, 1991;1998), Sissala (Blass, 1989;1990), Amharic and Swahili (Nicolle, 2000) and German (Lilley, 2013).
This study is based on a corpus of texts compiled using the PELCRA search engine of the National Corpus of Polish (Narodowy Korpus Języka Polskiego, NKJP, n.d.).A search was performed to find every example of jakoby between the years 2000 and 2010 in the following (mostly local and regional) media publications: 4 "Co tydzień Jaworzno", "Dziennik Łódzki", "Gazeta Goleniowska", "Gazeta Krakowska", "Gazeta Poznańska", "Gazeta Ubezpieczeniowa", "Gazeta Wrocławska", "Polityka", "Słowo Polskie -Gazeta Wrocławska" and "Trybuna Śląska".This yielded 835 examples in 752 texts.After all examples of jakoby in its hearsay function, defective examples, as well as any duplicates had been eliminated, 582 tokens remained for use in this study.An additional search was performed in the same publications over the same time period for examples of the inflected forms jakobym, jakobyś, jakobyśmy and jakobyście; this yielded twenty-four examples of jakobym and two of jakobyśmy, all "valid".The corpus thus contains a total of 608 examples.It is perhaps significant that of these 608 examples, 308 (51%) can be compared to example (2) above in that jakoby is embedded under a nominal that indicates that the accompanying proposition is attributed to someone other than the authorial voice.Examples include informacja ("information"), podejrzenia ("suspicions"), pogłoski ("rumours"), wnioski ("conclusions"), zarzuty ("accusations") and doniesienia ("reports").A further 176 examples (29%) are embedded under a verb of denial (example (3) above).In nearly all cases this is zaprzeczać/zaprzeczyć, although there are six instances of dementować/ zdementować and two cases of wypierać się/wyprzeć się (here "to deny").
Before we proceed to the main part of this paper it should be mentioned that jakoby is relatively low-frequency, generally occurring only in more formal written language.The PELCRA search engine of NKJP provides access to texts encompassing a total of just over 240 million words, in which there are 4168 examples of jakoby, in both its complementiser and hearsay functions; this corresponds to one occurrence of jakoby per approximately 57 thousand words.Assuming an overall complementiser-to-hearsay-particle ratio broadly similar to that of our own corpus (69.7% to 30.3%), we can estimate that, in written Polish, jakoby as a complementiser occurs around once in 83 thousand words.
The remainder of this paper is structured as follows.Section 2 provides a synopsis of relevance theory, focusing on metarepresentation.In section 3 we shall suggest how jakoby may be analysed from a relevance-theoretic point of view and make some observations regarding how it functions at the level of discourse.In section 4 we shall consider some anomalous occurrences of jakoby not readily accommodated within our analysis, and in section 5 we present our conclusions.

Relevance theory and metarepresentation
So far we have already made several references to indirect-speech-type constructions.According to relevance theory, such cases are metarepresentational because they involve "a representation of a representation: a higher-order representation with a lower-order representation embedded within it" (Wilson, 2012: 230).More specifically, the metarepresentational subordinate proposition is said to be an example of the interpretive use of language because it is seen as an interpretation of an attributed utterance or thought which it resembles in terms of its propositional content.Thus, in example (8) the embedded proposition (underlined) is understood as an interpretation of an utterance resembling ( 9 Że has been substituted for jakoby here because it is unmarked vis-à-vis jakoby and helps us discuss this example as an unremarkable instance of indirect speech. allowances have risen by 12% in relation to 2002, while the average allowance is just a little over 800 zł (the maximum permitted level is 1942 zł).") ("Gazeta Goleniowska", 13.12.2006)(9) Radni w Goleniowie uchwalili sobie gigantyczne podwyżki diet.
("Goleniów counsellors have awarded themselves gigantic allowance increases.")A crucial element of an utterance used interpretively is the writer-speaker's attitude towards it; this arises from the utterance's subordination to the writer-speaker's discursive purpose.Any attitude -from endorsement, through scepticism and doubt, to outright rejection -towards an attributed utterance may be implied contextually.The recovery of this attitude, like that of all speaker-intended meanings, proceeds according to the relevance-theoretic comprehension procedure.Relevance theory claims that every utterance "communicates a presumption of its own optimal relevance" (Sperber, Wilson, 1986/1995: 266-267).The presumption of optimal relevance entitles the hearer to presume that the speaker is expressing herself, to the extent that she is able, in such a way that the hearer is required to expend no gratuitous processing effort in recovering her intended meaning. 6nce a presumption of optimal relevance has been communicated to him by mere exposure to an utterance, the hearer tests interpretive hypotheses in order of accessibility.As soon as an interpretation is found that confirms the presumption of optimal relevance, he stops.
In ( 8) the adverbial w sensacyjnym tonie ("in a sensational tone") arguably signals the reporting voice's scepticism regarding the embedded claim.W rzeczywistości ("in reality") then suggests that what follows will contradict the previous proposition in some sense.Furthermore, the assumption that an increase of 12% over the previous four years and an average allowance of 800 zł are both relatively meagre is presumed to be highly accessible to the reader.These assumptions and contextual features work together to yield an optimally relevant interpretation on which the reporting voice is understood to have a dissociative attitude towards the attributed proposition; she herself does not consider the allowance increases to be "gigantic".
Relevance theorists also maintain that echoic utterances and verbal irony involve interpretive use.These have in common that the speaker expresses her own attitude -in the case of irony one of dissociation, such as "scepticism, mockery rejection etc." (Wilson, 2006(Wilson, : 1730) ) -towards an attributed utterance or thought.
Importantly, relevance theory recognises that negation can too be metarepresentational.Albu (2012a;2012b) suggests that "ordinary" or truth-conditional negation can either be descriptive or metarepresentational. Descriptive negation represents a state of affairs in terms of what it is not, which means that a "semantic affirmative correspondent" (SAC) is available; the same state of affairs can be described using either descriptive negation or the SAC.Imagine a group of friends playing a game of Taboo® and one of them has to describe the word "Pope" without saying "Catholic".She might say: (10) This person isn't a Protestant.

__ 169 __
Here, one of the available SACs is "This person is a Catholic".In the case of metarepresentational negation, the negated proposition is entertained as an interpretation -one on which it is rejected or denied -of an utterance or thought that is attributed to a person or persons relevant to the discourse (e.g. the hearer) in that any such persons are construed as (potentially) entertaining it.In other words, metarepresentational negation involves interpretive use.Albu terms the rejected proposition a "formal affirmative correspondent"; note that it plays no role in the interpretation of cases of descriptive negation.An example from Wilson (2012: 251): (11) PETER: Oh, you're in a miserable foul mood tonight.MARY: I'm not in a miserable foul mood tonight; I'm a little tired and would like to be left alone.Here, Peter inserts into the context the proposition "Mary is in a miserable foul mood tonight", which Mary subsequently rejects.Significantly, since metarepresentational negation can be conceived of in terms of denial, we can report Mary's negative utterance in ( 11) as ( 12), but reporting the statement in (10), i.e. descriptive negation, as ( 13) is, in the imagined context, decidedly odd: (12) Mary denied she was in a miserable foul mood.( 13) ?She denied this person was a Protestant.Some languages possess linguistic devices that make explicit that an utterance is to be understood as instance of interpretive use.Examples are Sissala rέ (Blass, 1989;1990), Japanese tte (Itani, 1991;1998), Amharic indɛ and Swahili je (Nicolle, 2000).The types of interpretive use that they indicate include reported speech and echoic uses.( 14), ( 15) and ( 16) are examples from Sissala (Blass, 1990); ( 14) is a reported-speech-type construction, in (15) rέ functions as a hearsay particle, while ( 16) is an echoic utterance where an attitude of endorsement is implied: (14) Ba sɛ rí ɓa yálá há kúé makɛ doƞ pinɛ weri pa wͻ.
("It is said that some died and were untied and left there.")(16) A: Ba dυla á wέri.
("They have done really well rέ.")Such devices are said to encode procedural meaning because they reduce the processing effort that utterance comprehension demands by constraining the inferential computations that the hearer is required to deploy (Blakemore, 1987;2002;Escandell-Vidal, Leonetti, Ahern, 2011;Wilson, Sperber, 2012;Carston, 2016, among others).Now that we have considered the theory that provides the framework for our investigation, we are in a position to begin considering the complementiser jakoby from a relevancetheoretic point of view.

Jakoby and metarepresentation
It emerges from our considerations on metarepresentation that those subordinate clauses where jakoby as a complementiser is licensed have in common that they are attributive, i.e., they involve interpretive use.A clear example is indirect speech: (17) Tekst opublikowany przez "Metro" w sensacyjnym tonie informuje, jakoby radni w Goleniowie uchwalili sobie gigantyczne podwyżki diet.
("A text published by 'Metro' informs readers in a sensational tone jakoby Goleniów counsellors have awarded themselves gigantic allowance increases.")("Gazeta Goleniowska", 13.12.2006)Closely related to indirect speech are cases where a nominal is accompanied by adnominal modifier that summarises an attributed proposition.In (18) the proposition introduced by jakoby is an interpretation of a stwierdzenie ("claim") that is attributed to the author of the article in which it was made: (18) Artykuł zawiera nieprawdziwe stwierdzenie, jakoby panowie kupili teren od miasta -kupili go od Gminy WIEJSKIEJ Głogów.
("The alleged crimes of the president consist in the notion jakoby he supposedly made permission to extend upwards dependent on obtaining the two flats by deception or by purchasing them -which is illegal -as well as the fact that he liked taking Porsches for a ride at the dealer's expense -which is a lesser crime.")("Polityka", no.2706, 23.05.2009)Clauses introduced by jakoby which are embedded under a negative matrix predicate are examples of metarepresentational negation because the negated proposition is understood as an interpretation (one on which it is rejected) of an utterance or thought attributed to a (construed) voice or voices somehow relevant to the discourse.In example (3) above, the matrix subject explicitly denies the claim that the fire concerned had earlier blazed near the headquarters; for this utterance to be felicitous as a denial this must previously have been at least suggested: (20) Zaprzeczył też, jakoby wcześniej ogień hulał jeszcze bliżej głównej siedziby spółki.
("It is untrue jakoby paper for passports was produced outside the country.")("Polityka", no.Here, the reader is construed as having inferred from the first sentence that modern Europe is Christian.This assumption -which is thus attributed to the reader -is then denied in the second sentence. Because of its association with attributed utterances and thoughts, we wish to suggest that jakoby, like Sissala rέ, Japanese tte, Amharic indɛ and Swahili je, encodes procedural meaning that makes explicit that a proposition is to be understood to be attributed.In many cases, such as examples (17) (indirect speech) and (18) (adnominal modifier), the nature of the matrix clause and head noun respectively indicate the attributive nature of the jakoby-clause, thus from a metarepresentational point of view jakoby is redundant.In other cases, jakoby makes a clear contribution in terms of metarepresentation; it ensures that the clause it introduces is understood to be attributed not to the author of the text, but to text-external voices (e.g.example ( 19)).Such an interpretation would arguably be inferable were że substituted for jakoby, but the recovery of such an interpretation would require greater processing effort.We also suggest that the constraint imposed by jakoby makes a contribution, albeit a subtle one, to cases of metarepresentational negation, such as examples ( 20), (21), and ( 22) above: it makes the metarepresentational nature of the denied proposition explicit.
Having established the metarepresentational nature of clauses introduced by jakoby, we return briefly to cases of hearsay jakoby, such as (6) above.In the examples above the higher-order representation under which the lower representation is embedded is stated explicitly in the form of the matrix clause (e.g.tekst […] informuje, or nieprawdą jest).However, in cases such as (6), repeated below, the procedural import of jakoby instructs the hearer-reader to construct an appropriate clause himself on the level of mental representation, something like ( 24

Jakoby and author attitudes
We have suggested that jakoby can be analysed as a marker of interpretive use that introduces attributed propositions; these include the denied or rejected proposition in the case of metarepresentational negation.We have also observed that in some cases, for example indirect speech, it is essentially redundant as a marker of metarepresentational use.This, however, is not to say that in such environments jakoby never plays a role.In this section we shall argue that an element of the procedural input of jakoby is to constrain the range of epistemic attitudes that it allows to be implied towards the embedded proposition.
The authors cited earlier in this paper hold that jakoby introduces an element of doubt or rejection to how the subordinate proposition is interpreted.In our view, this is justified and is consistent with the evidence of our corpus.We claim that the procedural import of jakoby is such that it explicitly distances the attributed proposition from the subjectivity of the reporting voice.In so doing, it blocks interpretations on which the reporting voice aligns herself with it, thereby narrowing the range of epistemic attitudes towards it that can be recovered inferentially by the hearer-reader.Nevertheless, there is still significant overlap between the attitudes that że and jakoby allow to be inferred contextually.Jakoby, however, is the marked term in the że-jakoby opposition, and, as such, encourages the recovery of attitudes that are further away from ones such as endorsement, which are compatible only with że; those attitudes constrained by jakoby therefore tend to be more readily associated with epistemic distance, such as scepticism, doubt and rejection.Thus agnostic interpretations are less likely to be recovered.However, they are not impossible, as evidenced by example (25): (25) Niezdementowany przeciek prasowy, jakoby strona polska była przeciwna wyborowi Joschki Fischera na europejskiego ministra spraw zagranicznych, potwierdza domniemanie, że również wobec Berlina dystans ciągle się utrzymuje.
("An undenied press leak [which says] jakoby Poland was opposed to Joschka Fischer's election as European foreign minister, is confirmed by the presumption that Berlin is still being held at a distance.")("Polityka", no.2413, 9.08.2003)Significantly, the leak is described as niezdementowany ('undenied'): if the leak has not been denied, then the possibility that it is true remains.Furthermore, we are subsequently told that this leak has been confirmed (albeit by a presumption).Despite the embedded proposition's likely being true, the reporting voice plausibly does not wish to align herself with it.Thus, we claim that she selects jakoby not in order to imply her doubts towards it, but to allow herself to remain impartial.
In (26), the procedural import of jakoby plays a perceptible role in encouraging the recovery of an interpretation on which the reporting voice entertains slight scepticism towards the embedded proposition: (26) Pierwsze diamenty wywodzą się z Indii, tam też (podobno) po raz pierwszy użyto terminu "karat".
("The first diamonds came from India, and it is there that the term 'carat' was (podobno) first used.There is also information jakoby the homeland of the carat was the Arab countries where the seed pods of the carob tree, known as 'keration', became an accepted unit of measure.However you look at it, the carat came into being thanks to… the seeds of plants.")("Gazeta Wrocławska", 29.08.2002)The last sentence indicates that the reporting voice allows for the possibility that the proposition introduced by jakoby -that the carat arose in Arab countries -is true.However, the choice of marked jakoby over że indicates that she entertains, albeit to a small degree, doubts regarding the second explanation.That she is more inclined towards the first one is also manifested in the opposition between jakoby and the bracketed hearsay particle podobno; the latter, unlike jakoby, is compatible with any evaluative stance of the reporting voice, including endorsement (Wiemer, 2006;Żabowska, 2008;Wiemer, Socka, 2017a;2017b).
In other cases, however, the attitude to be recovered towards an attributed proposition is made explicit by the co-text, thus the same interpretation would be available were że substituted for jakoby.In ( 27) the clause co jest nieprawdą ("which is untrue") makes explicit that the writer believes the content of the jakoby-clause to be untrue: (27) Firma CIGNA SA podaje jako uzasadnienie swojej decyzji, jakobym nienależycie zabezpieczyła pojazd przed kradzieżą, co jest nieprawdą.Wymagane zabezpieczenia były zamontowane w samochodzie, a przez cały czas zajścia znajdowałam się w bezpośredniej bliskości pojazdu.
("As justification for its decision the firm CIGNA SA states jakoby I have inadequately insured my vehicle against theft, which is untrue.The necessary safeguards were installed in the car and for the duration of the incident I remained in close proximity to the vehicle.")("Gazeta Ubezpieczeniowa", 29.06.2004)The occurrence of jakoby is particularly interesting in example ( 28 Hooper (1975) would classify as a semi-factive verb and can thus be said, in the normal case, to presuppose the truth of its embedded proposition.In (28), however, the final sentence indicates that the author of this letter to the editor does not believe the information that he 7 has learnt; dowiedzieć się is thus understood non-factively, and, furthermore, ironically.
Non-factive knowledge attributions from a relevance-theoretic point of view are the subject of Domaneschi and Di Paola (2019).Having used the results of an empirical study to show that "know has a literal meaning that is factive […] and that entails a truthcondition" while "[a]ll the non-factive uses of know are better considered as non-literal instances of the expression" (Domaneschi, Di Paola, 2019: 102), they go on to show that non-literal interpretations of "know" can be explained in terms ad hoc concepts, and, more specifically, loose use.
Relevance theory claims that ad hoc concept construction results from the pragmatic adjustment of the meaning of lexical items; this may involve either a narrowing or loosening (broadening) of the denotation, or literal meaning (Wilson, Carston, 2007).In the case of lexical narrowing, the hearer's expectation of optimal relevance results in an occurrence of the item with a meaning that is more restricted than its literal one.An example (from Carston, 2004: 642) is "He was upset but he wasn't upset" (uttered by a witness at the trial of a man accused of murdering his wife), where the reason why we do not see this as a contradiction is that we understand the first occurrence of "upset" as UPSET, i.e. its literal meaning, but the second occurrence we understand as a narrower concept UPSET*.According to Carston "[t]he second of the two concepts carries certain implications (e.g. that he was in a murdering state of mind) that the first one does not […]" (Carston, 2004: 642).
Like Domaneschi and Di Paolo in the case of non-factive occurrences of "know", we claim that dowiedzieć się in ( 28) is understood not as the literal DOWIEDZIEĆ SIĘ, but as an ad hoc concept characterised by a loosening of the literal meaning, i.e.DOWIEDZIEĆ SIĘ*.Here, the primary role of procedural jakoby is to force an interpretation of the embedded proposition on which it is understood as attributed to the writer of the quoted article.After all, the author of the letter, believing it to be false, wishes to distance himself from this proposition.Jakoby thus prompts a process of pragmatic adjustment that results in an optimally relevant interpretation whereby the sense "acquire information" is retained, but the requirement that the truth of this information be taken for granted is sacrificed.

7
The past-tense verb form dowiedziałem się indicates that the writer is male.This brings us to the end of our discussion of the attitudes that a writer may imply contextually towards a clause introduced by jakoby.In distancing a proposition explicitly from the subjectivity of the authorial voice, jakoby has the effect of encouraging the recovery of attitudes readily associated with epistemic distance, such as scepticism, doubt and rejection.Furthermore, in constraining interpretation in such a way that a proposition is understood to be attributed, it can force the construction of an ad hoc concept if the literal meaning of a lexical item is incompatible with an attributed subordinate proposition.

Global and local interpretations of jakoby
In the examples we considered in the last subsection, jakoby is selected from the point of view of the author of the text; it is the author who distances herself explicitly from the attributed proposition, and thus communicates an attitude such as rejection, doubt or scepticism towards it.We propose that in such cases jakoby receives a global reading because jakoby reflects the relationship, and subordination, of the embedded proposition to the macro-discursive purpose of the authorial voice.
In some cases, however, jakoby is selected from the perspective of the matrix subject, thus it is from her subjectivity that the procedural import of jakoby explicitly distances the embedded proposition; it is she who is construed as communicating an attitude compatible with jakoby towards this proposition.We suggest that jakoby is then accorded a local interpretation.In these cases, what is subordinated to the reporting voice's discursive purpose is not the proposition introduced by jakoby, but the matrix subject's interpretation of it.Perhaps, for example, she denies this proposition.This means that to the jakoby-proposition itself the reporting voice may imply contextually any evaluative stance, including ones on which she endorses it.
Perhaps curiously, our observation that jakoby may reflect the point of view of the matrix subject has apparently evaded previous authors on the subject.The only mention we have found is a footnote(!) in Wiemer (2006), who remarks: [i]t should be added that the negative epistemic evaluation can also be ascribed to [a] person different from the metaspeaker; cf. for instance: X sceptycznie podchodzi do zapewnień Y-a, jakoby P (X sceptically treats Y's assurances as if P).In such a case, it is the referent X named in the sentence who takes a sceptical stance towards another person's (Y's) assertions.(Wiemer, 2006: 44) Local interpretations occur when a jakoby-clause is embedded under a verb such as zaprzeczać/zaprzeczyć and dementować/zdementować ("to deny").Here, it is the matrix subject, rather than the authorial voice, who explicitly rejects a proposition that is understood to have been previously inserted into the context and thus to be attributed to a voice or voices other than the matrix subject: (29) Zaprzeczył też, jakoby wcześniej ogień hulał jeszcze bliżej głównej siedziby spółki.Nie zgadza się to z faktami.Reporterzy natrafili na duże i jeszcze gorące połacie wypalonej trawy właśnie w obrębie lotniska.
("He also denied jakoby fire had earlier been blazing even closer to the company's main headquarters.This does not tally with the facts.Reporters have come across large expanses of grass that was still burning within the territory of the airport.")("Gazeta Wrocławska", no.20/4, 2000) In this example [n]ie zgadza się to z faktami ("[t]his does not tally with the facts") makes explicit that the reporting voice, unlike the matrix subject, considers the (attributed) embedded proposition to be true; thus the former does not align herself with the position of the subject of the matrix clause.
Jakoby also reflects the point of view of the matrix subject, i.e. analysts, in (30).Here, the author cites these experts in order to address a perceived necessity to exclude from the discourse potential voices who, on the basis of the previous sentence, may have assumed a cause-effect relationship between later payment dates and lower incidence of bad debt: ( 30 ("Another girl from School no. 1, who was also hit with nerve gas, has been on the same ward for observation since Tuesday. […] Meanwhile, the headteacher has no idea jakoby an incident of this type happened on site on Tuesday.'It's true that I took some gas off one of the pupils, but I've heard nothing about one of the girls getting hit with it,' admits Bogusław Hrycyk.")("Trybuna Śląska", 24.10.2003)Here, jakoby constrains interpretation in such a way that the proposition we wtorek w placówce miało miejsce takie zdarzenie ("an incident of this type happened on Tuesday") is understood from Hrycyk's point of view as hearsay.After all, he knew nothing about the nerve gas incident until he was told about it; thus, at this point hearsay is the only evidence he has for the truth of the proposition.
The issue of global and local interpretations is also significant when the matrix clause to which the proposition introduced by jakoby is subordinate occurs itself within indirect speech.In these cases, the matrix subject is locally the reporting voice, therefore we might __ 177 __ expect jakoby to be selected from her point of view and thus to receive a local interpretation.An example where this is the case is ( 32 ("But Wright did not only offer a message of hope.After 11th September he called the attack God's punishment sent on America for the sins of slavery and racism as well as its imperial policy."May God condemn America!" he pontificated.He also suggested that he agreed with the theory jakoby the AIDS epidemic had been brought about in order to wipe out blacks.") ("Polityka", no. 2656("Polityka", no. , 31.05.2008)This also instantiates jakoby within a quotation, but this time the matrix subject (Wright) endorses the clause introduced by jakoby: he says he agrees with it.If the complementiser that introduces the proposition epidemię AIDS wyłowano w celu eksterminacji Murzynów ("the AIDS epidemic was brought about to wipe out blacks") had been chosen from the matrix subject's point of view then we would find że.However, this could be interpreted as ambiguous between a local and global reading; in the latter case the reporting voice could be accused of endorsing this highly controversial proposition.Instead, the reporting voice favours procedural jakoby over że.Since jakoby clashes epistemically with zgadza się z teorią ("agrees with the theory"), it forces a global interpretation that allows the authorial voice to distance herself from the relevant proposition.She therefore construes for herself an audience who she does not wish to believe that she aligns herself with it.
This provides us with some evidence that within indirect speech it is possible for the authorial voice to use the constraining effect of jakoby to exploit the distinction between global and local interpretations of this complementiser in order to distance herself from a position advanced.

Non-metarepresentational uses of jakoby: some counterexamples to our analysis
A comprehensive investigation of jakoby should not ignore cases that it is more difficult to explain in terms of the analysis that we have proposed.Wiemer (2015: 221-222) provides several examples where he believes that the clause that jakoby introduces is to be understood epistemically, rather than evidentially (or attributively, in our terms).Nevertheless, in our view Wiemer's examples can partially be accounted for in attributive (metarepresentational) terms.We would argue that in (34) the writer distances herself from a proposition that would be attributed to persons other than the writer if it were entertained: (34) Byłoby błędem sądzić, jakoby młodzież w wieku 18-17 lat nie kochała swojego miasta.
("The forefield of mutual relations has been demined, but the remnants of obstacles have not been cleared, which create the impression on some people jakoby the Polish-German war is still ongoing.")("Polityka", 24.09.2004)Nevertheless, it could also be argued that ( 36) and (37) (as well as the other three similar examples in our corpus) can be analysed as cases where an impression or feeling __ 180 __ Ultimately, cases of jakoby that cannot be analysed in metarepresentational terms appear to be marginal and for this reason we do not believe that their existence should be considered to pose a serious threat to our analysis.We suggest that they can perhaps be explained in terms of analogy with (metarepresentational) examples involving a jakoby-clause which is embedded under a negative predicate and is thus presented as false.

Conclusion
In order to account for its distribution, previous writers on jakoby, such as Wiemer (2006) and Stępień (2010a), have proposed both an evidential and an epistemic component for jakoby.We, however, have suggested that, with but a few exceptions, cases where jakoby occurs have in common that the embedded proposition is understood to be attributed.This includes not only propositions in the scope of jakoby as a hearsay particle and the embedded proposition of indirect-speech-type constructions, but also instances of metarepresentational negation.Thus, we have been able to provide a unified account, suggesting that in the normal case what instances of jakoby have in common is their metarepresentational nature.This prompts us to propose that jakoby can be added to the inventory of procedural items associated with interpretive uses of language, including Sissala rέ, Japanese tte, Amharic indɛ, Swahili je and the German reportative subjunctive.
As the marked term in the że-jakoby opposition, an element of its procedural effect is to distance the embedded proposition explicitly from the subjectivity of the authorial voice.In doing so, it defines the parameters of a semantic space which encompasses attitudes ranging from objectivity/neutrality to downright rejection, while blocking interpretations incompatible with epistemic distance, such as endorsement.Furthermore, we have suggested there are cases where the effect of embedding a jakoby-clause under a "factive" predicate, such as dowiadywać się/dowiedzieć się ("to find out") (example 28) is to force an attributive interpretation of this clause.As a result, the comprehension process is constrained in such a way that the hearer-reader understands the predicate as a "loose" use.
A further insight that this paper has developed is that jakoby can be selected from the perspective of the authorial voice herself (the reporting voice, in the case of indirect speech), which yields a global interpretation of jakoby, or from the point of view of the matrix subject.In the latter case a local interpretation results.
Further relevance-theoretic research should concern the Polish hearsay particles podobno, jakoby, rzekomo and ponoć, and consider how they instruct the hearer-reader to embed the proposition in their scope under a higher-order representation such as twierdzi się, że ("it is claimed that") (as suggested in section 3.1).Another Polish candidate for analysis as a marker of attributive use is the verb mieć ("to have") in its hearsay function, for example On miał ukraść samochód sąsiadowi ("He is said to have stolen his neighbour's car").

4
2010 is the last year for which texts are available.